CN108304810A - Aquatic bionic side line hydraulic pressure based on neural network and water flow field information detection method - Google Patents
Aquatic bionic side line hydraulic pressure based on neural network and water flow field information detection method Download PDFInfo
- Publication number
- CN108304810A CN108304810A CN201810123168.XA CN201810123168A CN108304810A CN 108304810 A CN108304810 A CN 108304810A CN 201810123168 A CN201810123168 A CN 201810123168A CN 108304810 A CN108304810 A CN 108304810A
- Authority
- CN
- China
- Prior art keywords
- hydraulic pressure
- neural network
- water flow
- side line
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Abstract
The aquatic bionic side line hydraulic pressure that the invention discloses a kind of based on neural network and water flow field information detection method, underwater hydraulic pressure flow information of water is acquired using MEMS sensor, hydraulic pressure water flow data corresponding to vibration source for different location establishes database, deep learning is carried out by deep neural network, training vibration source position distinguishes model, finally achieve the effect that the different vibration source positions of identification using trained model, realize the object recognition and detection ability of underwater robot, marine exploration is carried out for civilian and the army and the people's underwater robot, and a kind of new thinking and approach are provided.
Description
Technical field
The invention belongs to underwater environment discovery techniques fields, and in particular to a kind of aquatic bionic based on deep neural network
The detection method of side line hydraulic pressure and water flow field information Perception system.
Background technology
In recent years, the continuous attention with various countries to marine field, research and development and national defence for ocean it is waterborne
Security Construction is paid much attention to by domestic and foreign scholars, wherein a kind of novel marine engineering equipment of underwater robot, in military affairs
And marine exploration, underwater investigation etc. are used widely.And in these application fields.The identification of submarine target positions
The basis of its work and guarantee, therefore a kind of novel effective underwater information sensory perceptual system is urgently developed.
Fish and amphibian perceive underwater various ambient conditions by side line under water, therefore by bionic principle,
The perception that underwater hydraulic pressure flow is carried out using lateral-line system is significant, and at present to the bionical sensing of class neuromast of lateral line canal
Device systematic research is less, and the product of the type is also very few.
Invention content
In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is that providing a kind of based on depth
The aquatic bionic side line hydraulic pressure of neural network and method of the water flow field information Perception to detect vibration source position, by depth
The mode of habit identifies underwater vibration source position.
The present invention uses following technical scheme：
Aquatic bionic side line hydraulic pressure based on neural network and water flow field information detection method, are acquired using MEMS sensor
Underwater hydraulic pressure flow information of water, the hydraulic pressure water flow data corresponding to the vibration source for different location establish database, pass through depth
Neural network carries out deep learning, and training vibration source position distinguishes model, and it is different finally to reach identification using trained model
The effect of vibration source position realizes the object recognition and detection ability of underwater robot.
Specifically, including the following steps：
S1, according to the fairshaped underwater robot of fish bionic design, and determine its side line position；
S2, the side line location arrangements MEMS sensor determined along step S1；
S3, different location places vibration source around robot under water, and the MEMS sensor of step S2 arrangements is utilized to obtain
Hydraulic pressure corresponding to different location vibration source and water flow data；
S4, projected depth neural network, the hydraulic pressure water flow data that step S3 is obtained import in deep neural network, training
Provide the deep neural network model of identification vibration source position；
S5, vibration source position is derived using the step S4 deep neural network models established.
Further, in step S2, side line is located at underwater robot axis both sides symmetric position, and side line is provided on position
Pit for installing MEMS sensor.
Further, MEMS sensor can obtain hydraulic pressure pressure data and flow rate of water flow data simultaneously, export as binary channels
Spectrum information, and have water resistance.
Further, in step S3, the different location of different vibration sources is the difference with underwater robot same level
Coordinate position.
Further, a kind of building form of database is：90 differences are selected in front of bionic underwater robot head
Vibration source is placed in position, and 5 hydraulic pressure water flow datas of each station acquisition record the hydraulic pressure water-flow signal corresponding to different coordinates,
Build the database of 450 groups of hydraulic pressure water flow datas.
Further, another building form of database is：Underwater robot water proximate is divided into 36 regions,
10 positions are randomly selected in each area carries out hydraulic pressure water flow data acquisition, the data in the same region have a fling at one group,
The database of 36 classes totally 360 groups of hydraulic pressure water flow datas is obtained in total.
Further, in database, 75% training data as deep neural network is randomly selected, 15% for training
When test data, 15% for training after verify data.
Specifically, step S4 is specific as follows：
S401, using the collected hydraulic pressure water-flow signal frequency spectrum datas of step S3 as training sample, determine depth nerve net
The hidden layers numbers of algorithm；
S402, successively determine that the parameter value of every layer of denoising autocoder carries out successively pre- instruction using successively greedy algorithm
Practice, finally connecting one can be with the output layer with classification feature of output coordinate value；
S403, the position according to sample carry out the fine tuning of depth nerve net algorithm parameter using BP algorithm, determine depth god
Through network model parameter；
S404, the parameter determined according to step S403 complete the training of depth nerve net algorithm and export judging result.
Further, pre-training is specific as follows in step S402：
Using denoising autocoder as feature extraction, man made noise is fused to data during feature extraction
In, i-th of denoising autocoder of training initializes depth nerve net algorithm using i-th trained of denoising autocoder
I-th layer of hidden layers numbers；
If i≤N, N are the denoising autocoder number of plies in deep neural network, return, which restarts to train i-th, goes
Make an uproar autocoder DAE；
If i ＞ N, output layer is determined according to vibration source position.
Compared with prior art, the present invention at least has the advantages that：
The present invention is based on the aquatic bionic side line hydraulic pressure of neural network and water flow field information detection method, utilize a series of water
It presses flow composite MEMS (Micro-electromechanical System) sensor to form artificial lateral-line system and carries out data acquisition, corresponding to the vibration source for different location
Hydraulic pressure water flow data establishes database, the hydraulic pressure and flow information of water only perceived by underwater robot side line, not against it
His specific condition, improves efficiency and the intelligence of Underwater Targets Recognition, passes through deep neural network and carries out deep learning, training
Vibration source position distinguishes model, finally achievees the effect that the different vibration source positions of identification using trained model, realizes underwater
The object recognition and detection ability of robot, for civilian and the army and the people's underwater robot carry out marine exploration provide a kind of new thinking and
Approach.
Further, side line is located at underwater robot axis both sides symmetric position, improves the positioning of lateral-line system target
Precision and range；It is twin-channel frequency spectrum that MEMS sensor can obtain hydraulic pressure pressure data and flow rate of water flow data and export simultaneously
Information carries out prediction vibration source target location using two kinds of data of hydraulic pressure and flow, further improves precision of prediction.
Further, the different location of vibration source is located at same level with underwater robot in database so that data
Data in library have sustained height location information with tested vibrating body, and study can be reduced when being predicted by neural network
Intensity.
Further, a kind of building form of database is that 90 different positions are selected in front of bionic underwater robot head
Placement location vibration source builds 450 groups of hydraulic pressure water flow datas, is trained test to neural network by large sample, can improve pre-
Precision is surveyed, and prevents the generation of over-fitting and poor fitting phenomenon.
Further, another building form of database is that underwater robot water proximate is divided into 36 regions to obtain
36 classes totally 360 groups of hydraulic pressure water flow datas are obtained, test is trained to neural network by large sample, precision of prediction can be improved,
And prevent the generation of over-fitting and poor fitting phenomenon.
Further, database randomly selects 75% training data as deep neural network, and 15% for when training
Test data, 15% for training after verify data, tested with mass data (75%), poor fitting will be prevented
Phenomenon generates, while being tested with 15% data while training deep neural network, and the production of over-fitting will be prevented
It is raw, it is finally verified with verify data (15%), will examine and ensure the final prediction output accuracy of deep neural network.
Further, the deep neural network with identification vibration source position is trained using the deep neural network of design
Model, by information collection, feature extraction, target prodiction output and etc. integration, in the case where ensureing precision simplification
The message processing flow of Underwater Targets Recognition.
Further, the training process of deep neural network is divided into the unsupervised pre-training to multilayer denoising automatic coding machine
Two steps of fine tuning are supervised with having for final prediction output, the cumbersome mistake for carrying out forming label to database data can be reduced
Journey, while the training time of neural network can be reduced.
Below by drawings and examples, technical scheme of the present invention will be described in further detail.
Description of the drawings
Fig. 1 is that the present invention is based on the aquatic bionic side line hydraulic pressure of deep neural network and water flow field information detecting method flow
Schematic diagram；
Fig. 2 is side line hydraulic pressure water flow sensing unit system arrangement schematic diagram of the present invention；
Fig. 3 is the structural schematic diagram that deep neural network of the present invention trains flow using data；
Fig. 4 is different vibration source position structural schematic diagrams in database of the present invention；
Fig. 5 is another different vibration source position structural schematic diagrams in database of the present invention；
Fig. 6 is expanded application structural schematic diagram of the present invention.
Specific implementation mode
The aquatic bionic side line hydraulic pressure that the present invention provides a kind of based on deep neural network and water flow field information Perception system
The detection method of system, using the underwater hydraulic pressure flow information of water of MEMS sensor acquisition, corresponding to the vibration source for different location
Hydraulic pressure water flow data establishes database, carries out deep learning by deep neural network, training vibration source position distinguishes model, most
Achieve the effect that the different vibration source positions of identification using trained model eventually, realizes the object recognition and detection energy of underwater robot
Power.
Referring to Fig. 1, a kind of aquatic bionic side line hydraulic pressure based on deep neural network of the invention and water flow field information sense
The detection method for knowing system, includes the following steps：
S1, according to the fairshaped underwater robot of fish bionic design, and determine its side line position；
The arrangement that the hydraulic pressure water flow sensing unit of side line shape is designed using bionical fish streamlined structure, can be more true
Acquisition subsurface water current information.
S2, sensor arrangement is carried out along the side line；
By arranging MEMS sensor in fairshaped side line structure, to simulate fish and the underwater sense of amphibian
Know that system, MEMS sensor can obtain hydraulic pressure pressure data and flow rate of water flow data simultaneously, exports as twin-channel spectrum information；
Secondly, MEMS sensor used has water resistance, can directly be measured under water in exposure；
Side line is located at underwater robot axis both sides symmetric position；It is provided on side line for installing hydraulic pressure water flow sensing unit
Pit, as shown in Figure 2.
The number of probes arranged is more, then the hydraulic pressure water flow data obtained is more accurate, but sensor is excessively intensive
It can then cause to interfere hydraulic pressure water flow data between sensor.
The arrangement number of sensor and interval are subject to data between sensor and are not interfered with each other in the present embodiment.
S3, different location places vibration source around robot under water, utilizes the hydraulic pressure water flow sensing unit to obtain different
Hydraulic pressure corresponding to the vibration source of position and water flow data；
The different location of different vibration sources refers to coordinate positions different from underwater robot same level, shakes to establish
The hydraulic pressure water flow data library of dynamic source different location, needs vibration source being positioned over different location, 5 hydraulic pressure water of each station acquisition
Flow data selects 90 different locations, as shown in figure 4, recording corresponding to different coordinates in front of bionic underwater robot head
Hydraulic pressure water-flow signal, build 450 groups of data database.
Another kind establishes the mode in the hydraulic pressure water flow data library of vibration source different location as shown in figure 5, by underwater robot
Water proximate is divided into 36 regions, randomly selects 10 positions in each area and carries out hydraulic pressure water flow data acquisition, same
Data in a region have a fling at one group, obtain the database of 36 classes totally 360 groups of data in total.
In database, 75% training data as deep neural network is randomly selected, 15% for test when training
Data, 15% is used for the verify data after training.
S4, projected depth neural network import the hydraulic pressure water flow data in the deep neural network, and training is provided
There is the deep neural network model of identification vibration source position；
Referring to Fig. 3, being in good experimental trough ring by the collected hydraulic pressure of step S3 institutes, water-flow signal frequency spectrum data
It is carried out under border, and there are many uncontrollable factors under actual complex oceanographic condition, the interference of various ocean object vibration sources is made an uproar
Sound will produce a large amount of useless feature, based on deep learning by the learning process of simulation brain, builds profound model and comes
The feature implied in learning data finally promotes accuracy of identification to portray the internal information of data rich；
Using hydraulic pressure, water-flow signal frequency spectrum as training sample, the hidden layers numbers of depth nerve net algorithm (DNN) are determined, it is deep
The training of degree neural network includes pre-training and fine tuning two parts, and pre-training successively determines every layer of denoising using successively greedy algorithm
The parameter value of autocoder, finally connects output layer；
Deep neural network is used as feature extraction using denoising autocoder (DAE), will during feature extraction
Man made noise is fused in data, improves robustness of the neural network in terms of feature extraction, and i-th of DAE of training uses training
I-th of DAE initialization depth nerve net algorithm (DNN) i-th layer of hidden layers numbers, N be denoising autocoder the number of plies, such as
Fruit i≤N, return restarts i-th of denoising autocoder DAE of training, if i ＞ N, is determined and is exported according to vibration source position
Layer；
The fine tuning that overall network parameter is carried out using BP algorithm determines deep neural network model parameter, completes depth god
Through net algorithm (DNN) training and export judging result.
BP algorithm is made of learning process the forward-propagating of signal and two processes of backpropagation of error.Due to more
The training of layer feedforward network is through frequently with error backpropagation algorithm.
S5, the deep neural network model is utilized, you can derive vibration source position from hydraulic pressure flow information of water.
Complete deep neural network after training, the hydraulic pressure water flow data that underwater robot is acquired, by lateral line system
Input of the hydraulic pressure flow information of water as deep neural network model on system measured by each sensor, passes through neural network multilayer
The feature extraction of denoising autocoder and the coordinate output of final output layer, you can complete the fixation and recognition of underwater vibration source.
Referring to Fig. 6, by the way that the output layer information of deep neural network is become three-dimensional from two dimension, i.e., the two of script are flat
Areal coordinate information becomes three-dimensional information as shown in the figure (L, φ, θ), and wherein L indicates vibration source to the distance of machine fish, θ expression lines
Section L indicates the angle of line segment L and x/y plane in the projection of x/y plane and the angle of y-axis, φ.Change deep neural network mould simultaneously
Type and the data volume for expanding database, i.e., also extend to three dimensions by the selection of coordinate points in database by two dimensional surface, can
Applied to the vibration source position in detection of three dimensional space.
The above content is merely illustrative of the invention's technical idea, and protection scope of the present invention cannot be limited with this, every to press
According to technological thought proposed by the present invention, any change done on the basis of technical solution each falls within claims of the present invention
Protection domain within.
Claims (10)
1. the aquatic bionic side line hydraulic pressure based on neural network and water flow field information detection method, which is characterized in that utilize MEMS
Sensor acquires underwater hydraulic pressure flow information of water, and the hydraulic pressure water flow data corresponding to the vibration source for different location establishes data
Library carries out deep learning by deep neural network, and training vibration source position is distinguished model, finally reached using trained model
To the effect for recognizing different vibration source positions, the object recognition and detection ability of underwater robot is realized.
2. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 1 and water flow field information detection side
Method, which is characterized in that include the following steps：
S1, according to the fairshaped underwater robot of fish bionic design, and determine its side line position；
S2, the side line location arrangements MEMS sensor determined along step S1；
S3, different location places vibration source around robot under water, utilizes the MEMS sensor of step S2 arrangements to obtain different
Hydraulic pressure corresponding to the vibration source of position and water flow data；
S4, projected depth neural network, the hydraulic pressure water flow data that step S3 is obtained import in deep neural network, and training is provided
There is the deep neural network model of identification vibration source position；
S5, vibration source position is derived using the step S4 deep neural network models established.
3. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 2 and water flow field information detection side
Method, which is characterized in that in step S2, side line is located at underwater robot axis both sides symmetric position, is arranged on side line position useful
In the pit of installation MEMS sensor.
4. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 3 and water flow field information detection side
Method, which is characterized in that MEMS sensor can obtain hydraulic pressure pressure data and flow rate of water flow data simultaneously, export as twin-channel frequency
Spectrum information, and there is water resistance.
5. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 2 and water flow field information detection side
Method, which is characterized in that in step S3, the different location of different vibration sources is coordinate different from underwater robot same level
Position.
6. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 5 and water flow field information detection side
Method, which is characterized in that a kind of building form of database is：90 different locations are selected in front of bionic underwater robot head
Vibration source is placed, 5 hydraulic pressure water flow datas of each station acquisition record the hydraulic pressure water-flow signal corresponding to different coordinates, build
The database of 450 groups of hydraulic pressure water flow datas.
7. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 5 and water flow field information detection side
Method, which is characterized in that another building form of database is：Underwater robot water proximate is divided into 36 regions,
10 positions are randomly selected in each region and carry out hydraulic pressure water flow data acquisition, and the data in the same region have a fling at one group, always
The database of 36 classes totally 360 groups of hydraulic pressure water flow datas is obtained altogether.
8. a kind of aquatic bionic side line hydraulic pressure based on neural network described according to claim 6 or 7 is visited with water flow field information
Survey method, which is characterized in that in database, randomly select 75% training data as deep neural network, 15% for instructing
Test data when practicing, 15% is used for the verify data after training.
9. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 1 and water flow field information detection side
Method, which is characterized in that step S4 is specific as follows：
S401, using the collected hydraulic pressure water-flow signal frequency spectrum datas of step S3 as training sample, determine depth nerve net algorithm
Hidden layers numbers；
S402, successively determine that the parameter value of every layer of denoising autocoder carries out successively pre-training using successively greedy algorithm, most
Connecting one afterwards can be with the output layer with classification feature of output coordinate value；
S403, the position according to sample are carried out the fine tuning of depth nerve net algorithm parameter using BP algorithm, determine depth nerve net
Network model parameter；
S404, the parameter determined according to step S403 complete the training of depth nerve net algorithm and export judging result.
10. a kind of aquatic bionic side line hydraulic pressure based on neural network according to claim 9 and water flow field information detection
Method, which is characterized in that pre-training is specific as follows in step S402：
Using denoising autocoder as feature extraction, man made noise is fused in data during feature extraction,
I-th of denoising autocoder of training initializes the of depth nerve net algorithm using trained i-th of denoising autocoder
I layers of hidden layers numbers；
If i≤N, N are the denoising autocoder number of plies in deep neural network, return restarts i-th of denoising of training certainly
Dynamic encoder DAE；
If i ＞ N, output layer is determined according to vibration source position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810123168.XA CN108304810B (en) | 2018-02-07 | 2018-02-07 | Underwater bionic lateral line water pressure and water flow field information detection method based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810123168.XA CN108304810B (en) | 2018-02-07 | 2018-02-07 | Underwater bionic lateral line water pressure and water flow field information detection method based on neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108304810A true CN108304810A (en) | 2018-07-20 |
| CN108304810B CN108304810B (en) | 2021-05-28 |
Family
ID=62864643
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810123168.XA Active CN108304810B (en) | 2018-02-07 | 2018-02-07 | Underwater bionic lateral line water pressure and water flow field information detection method based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108304810B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109238245A (en) * | 2018-11-30 | 2019-01-18 | 中国海洋大学 | A kind of novel bionic side line sensor |
| CN109298430A (en) * | 2018-08-08 | 2019-02-01 | 西安交通大学 | A kind of underwater composite bionic detection device and detection information fusion method |
| CN109902617A (en) * | 2019-02-25 | 2019-06-18 | 百度在线网络技术（北京）有限公司 | A kind of image identification method, device, computer equipment and medium |
| CN110065607A (en) * | 2019-05-17 | 2019-07-30 | 中国科学院自动化研究所 | Assist bionic machine fish |
| CN110763428A (en) * | 2019-10-16 | 2020-02-07 | 北京机电工程研究所 | Sensor layout method for sensing flow field around bionic fish body |
| CN112147712A (en) * | 2020-09-28 | 2020-12-29 | 中国海洋大学 | Underwater vibration source detection device and method |
| CN114354082A (en) * | 2022-03-18 | 2022-04-15 | 山东科技大学 | Intelligent tracking system and method for submarine pipeline based on imitated sturgeon whiskers |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105333988A (en) * | 2015-11-25 | 2016-02-17 | 中国海洋大学 | Artificial lateral line pressure detection method |
| CN106564577A (en) * | 2016-11-02 | 2017-04-19 | 中国海洋大学 | Multifunctional AUV based on bionic lateral line |
-
2018
- 2018-02-07 CN CN201810123168.XA patent/CN108304810B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105333988A (en) * | 2015-11-25 | 2016-02-17 | 中国海洋大学 | Artificial lateral line pressure detection method |
| CN106564577A (en) * | 2016-11-02 | 2017-04-19 | 中国海洋大学 | Multifunctional AUV based on bionic lateral line |
Non-Patent Citations (2)
| Title |
|---|
| JIEXIONG TANG等: "Compressed-Domain Ship Detection on Spaceborne", 《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》 * |
| YUECHAO CHEN等: "The research of underwater target recognition method based on deep learning", 《2017 IEEE INTERNATIONAL CONFERENCE ON SIGNAL PROCESSING, COMMUNICATIONS AND COMPUTING (ICSPCC)》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109298430A (en) * | 2018-08-08 | 2019-02-01 | 西安交通大学 | A kind of underwater composite bionic detection device and detection information fusion method |
| CN109298430B (en) * | 2018-08-08 | 2020-10-27 | 西安交通大学 | Underwater composite bionic detection device and detection target fusion identification method |
| CN109238245A (en) * | 2018-11-30 | 2019-01-18 | 中国海洋大学 | A kind of novel bionic side line sensor |
| CN109902617A (en) * | 2019-02-25 | 2019-06-18 | 百度在线网络技术（北京）有限公司 | A kind of image identification method, device, computer equipment and medium |
| CN110065607A (en) * | 2019-05-17 | 2019-07-30 | 中国科学院自动化研究所 | Assist bionic machine fish |
| CN110763428A (en) * | 2019-10-16 | 2020-02-07 | 北京机电工程研究所 | Sensor layout method for sensing flow field around bionic fish body |
| CN110763428B (en) * | 2019-10-16 | 2021-05-11 | 北京机电工程研究所 | Sensor layout method for sensing flow field around bionic fish body |
| CN112147712A (en) * | 2020-09-28 | 2020-12-29 | 中国海洋大学 | Underwater vibration source detection device and method |
| CN112147712B (en) * | 2020-09-28 | 2021-10-01 | 中国海洋大学 | Underwater vibration source detection device and method |
| CN114354082A (en) * | 2022-03-18 | 2022-04-15 | 山东科技大学 | Intelligent tracking system and method for submarine pipeline based on imitated sturgeon whiskers |
| CN114354082B (en) * | 2022-03-18 | 2022-05-31 | 山东科技大学 | Intelligent tracking system and method for submarine pipeline based on imitated sturgeon whisker |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108304810B (en) | 2021-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108304810A (en) | Aquatic bionic side line hydraulic pressure based on neural network and water flow field information detection method | |
| CN105333988B (en) | A kind of artificial side line pressure detection method | |
| CN112163461B (en) | Underwater target identification method based on multi-mode fusion | |
| CN103971406B (en) | Submarine target three-dimensional rebuilding method based on line-structured light | |
| CN109558859B (en) | Mining area distribution information extraction method and system based on DInSAR and DCNN | |
| CN105196114B (en) | Tool wear realtime on-line monitoring method based on wavelet analysis and neutral net | |
| CN109100741A (en) | A kind of object detection method based on 3D laser radar and image data | |
| CN105528585B (en) | Underwater vehicle detecting method based on ship wake and submarine topography internal wave models | |
| CN102306301A (en) | Motion identification system by simulating spiking neuron of primary visual cortex | |
| CN109655815A (en) | Sonar target detection method based on SSD | |
| CN117198330B (en) | Sound source identification method and system and electronic equipment | |
| CN112644646A (en) | Underwater robot intelligent system for large-water-area fish resource investigation and working method | |
| CN113420440B (en) | Vortex identification method based on ocean vertical structure | |
| CN109613623B (en) | Lithology prediction method based on residual error network | |
| CN112986914A (en) | Individual helmet and target sound source positioning and voiceprint recognition method thereof | |
| CN111637874A (en) | Multi-AUV layered detection system and detection method for red tide sea area | |
| CN115097534A (en) | Method and system for identifying space morphology of dry hot rock fracturing transformation | |
| Yu et al. | Underwater targets recognition based on multiple AUVs cooperative via recurrent transfer-adaptive learning (RTAL) | |
| CN113901927B (en) | Underwater object shape recognition method based on flow field pressure time course | |
| Połap et al. | Bilinear pooling with poisoning detection module for automatic side scan sonar data analysis | |
| CN110533650A (en) | A kind of AUV submarine pipeline detecting and tracking method of view-based access control model | |
| CN109991670A (en) | A kind of submarine target localization method | |
| CN109726637A (en) | The parking site detecting system and method in parking stall region are indicated with resilient polygonal | |
| CN114354082B (en) | Intelligent tracking system and method for submarine pipeline based on imitated sturgeon whisker | |
| CN114662603A (en) | Ship-borne multivariate marine environment information acquisition visualization algorithm |
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 |