CN108960421A - The unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network of improvement - Google Patents
The unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network of improvement Download PDFInfo
- Publication number
- CN108960421A CN108960421A CN201810577099.XA CN201810577099A CN108960421A CN 108960421 A CN108960421 A CN 108960421A CN 201810577099 A CN201810577099 A CN 201810577099A CN 108960421 A CN108960421 A CN 108960421A
- Authority
- CN
- China
- Prior art keywords
- ship
- neural network
- surface vehicle
- vehicle speed
- plane
- 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
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Feedback Control In General (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network.Data are collected, the four systems index influential on predetermined speed of needs is picked out；The four systems index is identified and handled and the nondimensionalization of all indexs；Principal component analysis is carried out to the system achievement data after four nondimensionalizations；Unmanned surface vehicle speed of a ship or plane prediction BP neural network is initialized；Network is trained with four systems index sample set；It tests to the generalization ability of unmanned surface vehicle speed of a ship or plane prediction BP neural network, is analyzed and corrected；BP neural network is predicted by the revised unmanned surface vehicle speed of a ship or plane, obtains the speed of subsequent time unmanned boat.The forecasting procedure of the speed of a ship or plane of unmanned surface vehicle provided by the invention is clear in structure, and logicality is stronger, is easy to write computer program realization.The present invention is suitable for the prediction of the unmanned surface vehicle speed of a ship or plane and trajectory planning, sea avoidance aspect.
Description
Technical field
The present invention relates to a kind of speed of a ship or plane prediction technique, specifically a kind of unmanned surface vehicle speed of a ship or plane prediction technique.
Background technique
Unmanned surface vehicle is the intelligent platform in surface navigation, and the aquatic environment as locating for unmanned surface vehicle is very multiple
It is miscellaneous, so the speed monitoring of unmanned boat is particularly significant.Since the measuring instrument (GPS) that unmanned boat carries may be blocked in short term
, the problems such as measurement noise caused by environmental perturbation is big, it is therefore desirable to the speed of unmanned boat under specific situation is predicted, is
Path planning and speed of a ship or plane control provide velocity prediction etc..Therefore, how to improve the prediction of the unmanned surface vehicle speed of a ship or plane is present research heat
Point, in the document that oneself has, many scholars by wavelet transformation, Kalman filtering and the methods of it is pre- to the speed of unmanned surface vehicle
Survey is studied.But these methods are all based on model, it is very low to the adaptivity of environment.
Dalian University of Technology's clock it is loud and clear research Large Container Ship the speed of a ship or plane forecast, Ship's Principal Dimensions, form coefficient and
After main engine power is selected, it is related to the speed of a ship or plane of ship using BP neural network prediction, passes through input shipping data, debugging training first
Network model obtains weight and threshold value, then will design in the desired scale input network of ship, and obtain predicted value.The disadvantage is that depositing
Reach a very big value in the number that certain probability works as training, but there are no meet the requirements for precision.
Harbin Engineering University Zhang Mingjun etc. is had using improved Elman network as multi-step Predictive Model
The prediction of speed and control experiment of Neural Network Online learning functionality and the underwater robot without on-line study function, and just
PREDICTIVE CONTROL effect has carried out comparative analysis.Deficiency is if underwater robot payload, underwater operation environment etc. occur
Change causes its dynamic characteristic to change, and lacks stronger adaptive ability.
The velocity information of Chinese Academy of Sciences's Huang target in order to obtain such as plum forever, it is proposed that Kalman's filter of target velocity prediction
Wave method, predicts the velocity information of tracked target, to constitute complex controll.Emulation and the experimental results showed that, by upper
The method of stating can be derived that accurate target speed information, but if in the case that the velocity information for introducing target is inaccurate,
System is easy diverging, causes to predict target information lag and precision is very low.
Beijing Jiaotong University Gong Shan is based on floating car data, in conjunction with the history run and real time traffic data of road network, establishes
One neighbor prediction model, the neighbour based on clustering and effectively solved determine that method will be to traffic information in grey forecasting model
State is divided, and is imported and carried out forecast analysis in model.But do not fully consider influence of the intersection to running speed also, it is right
In the processing methods of intersection data, there is still a need for studied.
Summary of the invention
It can be mentioned to avoid the influence of human factor, for path planning and speed of a ship or plane control the purpose of the present invention is to provide a kind of
For velocity prediction improvement based on the unmanned surface vehicle speed of a ship or plane online forecasting method of BP neural network.
The object of the present invention is achieved like this:
(1) data are collected, the parameter influential on predetermined speed of needs is picked out, specifically includes preceding 2000 groups of propulsion
The revolving speed of device, the wind speed and direction of sea wind, ocean current flow velocity and flow direction, the wave height of wave and wavelength and wave direction four systems refer to
Mark；
(2) the four systems index is identified and is handled and the nondimensionalization of all indexs；
(3) principal component analysis is carried out to the system achievement data after four nondimensionalizations；
(4) unmanned surface vehicle speed of a ship or plane prediction BP neural network is initialized；
(5) network is trained with four systems index sample set；
(6) it tests to the generalization ability of unmanned surface vehicle speed of a ship or plane prediction BP neural network, is analyzed and repaired
Just；
(7) BP neural network is predicted by the revised unmanned surface vehicle speed of a ship or plane, obtains the speed of subsequent time unmanned boat.
The present invention may also include:
1. training is grouped using repetitive exercise mode on the parameter influential on predetermined speed for picking out needs, it is first
First dividing six training datas is one group, is trained to BP neural network, predicts the 7th output data, then by nearest survey
Sample example is trained together with next group of data as training data.
It being identified described in 2. and is handled and the nondimensionalization of all indexs specifically includes: being handled using equalization method
Each achievement data of data set calculates the Eigen Covariance battle array R=cov (X) of primary standard sample set, asks that covariance matrix R's is complete
Portion's characteristic value and corresponding feature vector, and be ranked up according to characteristic value by substantially small.
3. carry out principal component analysis described in specifically includes: using the Principal Component Analysis in Multielement statistical analysis method
Multiple variables are selected significant variable by linear transformation by PCA, obtain variance by the principal component of substantially small sequence, wherein variance
Size be characteristic value, principal component analysis is carried out to each characteristic value and if weight E < 0.85 casts out this feature value；If power
Weight E >=0.85, then this feature value is chosen, and calculate the variance accumulation serial number of this feature value.
4. BP neural network, which carries out initialization and specifically includes, to be predicted to the unmanned surface vehicle speed of a ship or plane described in: establishing one with function
A trainable BP feedforward network, function need four input parameters, and first parameter is the matrix of a R*2 to define R
The maximum value and minimum value of input vector；Second parameter is the array of every layer of neuron number of a network；Third parameter
It is the cell array comprising every layer of transfer function title used；The last one parameter is the title for the training function used.
5. the generalization ability to unmanned surface vehicle speed of a ship or plane prediction BP neural network is tested, is analyzed and be subject to
Amendment specifically includes: the data result of prediction is compared with real data, ask RMSE be used to measure observation with true value it
Between deviation root-mean-square errorWherein x is average, N is number of samples, examines knot
Fruit casts out the prediction result if RMSE < 60%, is grouped again to training set data, strengthens the training of neural network；If
RMSE >=60% prove that the prediction effect of the experiment reaches aspiration level, export the prediction result of the training group.
It is specifically a kind of using BP neural network the present invention provides a kind of unmanned surface vehicle speed of a ship or plane prediction technique
Unmanned surface vehicle speed of a ship or plane prediction technique, during unmanned surface vehicle navigation, according to the revolving speed of a period of time propeller, sea wind
(wind speed/wind direction), ocean current (flow velocity/flow direction), wave (wave height/wavelength/wave direction) four systems index, can predict subsequent time
The speed of a ship or plane of unmanned surface vehicle.This method is suitable for the prediction of the unmanned surface vehicle speed of a ship or plane and trajectory planning, sea avoidance aspect, according to calculation
Using training set and forecast set data separating history instruction is added automatically after prediction in forecast set data by the principle of method iteration
Practice collection, the BP neural network that can timely update model improves model to the robustness of the variation of environment, has to its practical application
Very big significance.
The main feature of unmanned surface vehicle speed of a ship or plane forecasting procedure of the invention is embodied in:
(1) BP reverse transmittance nerve network algorithm is used, is a kind of multilayer feedforword net by Back Propagation Algorithm training
Network is one of current most widely used neural network model.BP network can learn and store a large amount of input-output mode and reflect
Relationship is penetrated, without obtaining the math equation of this mapping relations in advance.Training can all feed back neural network every time, pass through
Training after crossing feedback promotes higher accuracy rate, realizes more accurate prediction effect.
(2) different from the training method of traditional BP reverse transmittance nerve network algorithm, using repetitive exercise method, every six numbers
According to being one group, next data are predicted.New training group is added in the data group trained by next round training, will be nearest
Test sample as training examples expand training set, be equivalent to and in a disguised form increase data training set, improve neural network
Accuracy.
(3) training set is analyzed using Principal Component Analysis, wherein each principal component can reflect original variable
Most information, and information contained does not repeat mutually.In the prediction of speed experimental study of unmanned boat, to data set carry out it is main at
Analysis is screened out from it the higher index of weight accounting for influencing unmanned surface vehicle travel speed, has filtered out respectively previous
Revolving speed, sea wind (wind speed/wind direction), ocean current (flow velocity/flow direction), wave (wave height/wavelength/wave direction) letter of the propeller of section time
Breath, reduces the training dimension of data set, improves the speed of service and accuracy rate of program.
The present invention is based on traditional BP neural network models, have fully considered the environmental information of unmanned surface vehicle work, can
According to the data inner link that unmanned surface vehicle obtains, there can be good comprehensive treatment capability and good non-to gibberish
Linear approximation ability and the characteristics such as adaptive, are able to solve influence that measuring instrument (GPS) is blocked in short term and environmental perturbation is led
The measurement noise of cause causes that the accurate velocity information of unmanned boat can not be obtained in a short time greatly, provides for path planning and speed of a ship or plane control
Velocity prediction etc..Therefore, contain relationship and mathematical model without setting up the aobvious of complicated nonlinear system, it can be to avoid many artificial
The influence of factor, the difficulty that also can solve many limitations of traditional prediction method and face.
Detailed description of the invention
Fig. 1 is the flow chart for improving the unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network.
Fig. 2 is to increase data training set schematic diagram.
Fig. 3 is principal component neural network diagram.
Fig. 4 is BP neural network physical training condition.
Fig. 5 is the input of BP neural network node, output schematic diagram.
Specific embodiment
It illustrates below and the present invention is described in more detail.
The present invention is to provide a kind of unmanned surface vehicle speed of a ship or plane prediction technique using BP neural network, for the water surface nobody
The real-time prediction future time speed of a ship or plane during ship navigation, provides important reference for unmanned surface vehicle control decision.BP nerve
Network is broadly divided into two processes.First, working signal forward direction transmits subprocess；Second, error signal back transfer subprocess.
One three layers of BP network can complete the mapping of arbitrary M dimension to N-dimensional.Wherein, first layer is input layer, and the second layer is that have
The hidden layer of uncertain node number, third layer is output layer.Each layer of neurode is all inputted by preceding layer.And network section
The output of point can be illustrated in fig. 5 shown below as next layer of input or the output of whole network.
In BP neural network, only hidden layer node number is uncertain, and the node number of input layer and output layer is all
Determining, and how much meetings of hidden layer node number have an impact neural network.In fact, hidden layer node number can
To be determined by an empirical equation, it may be assumed that
Wherein, h is hidden layer node number, and m is input layer number, and n is output layer interstitial content, a be 1 to 10 it
Between regulating constant.
The Mathematical representation of BP neural network is as follows:
(1) output of hidden layer (BIAS is considered as another input node)
(2) calculating of output layer
H (k)=* s (j) O (k)=f (h (k))；
Wherein ii --- the input of network,
The output of O (k) --- network,
Wij --- the connection weight between the 1st layer of node i and the 2nd layer of node j is represented,
Wjk --- the connection weight between the 2nd layer of node j and the 3rd layer of node k is represented,
The transfer function of f (x) --- this node such as sigmoid:f (x)=training purpose is to reach error most
It is small, it is defined as follows:
Wherein
K-th of the output of o (p, k) --- input sample p,
K-th of target of t (p, k) --- input sample p exports
Neural network algorithm is a kind of neural network of supervised study, and " training " is to adjust weight, it is ensured that
Desired output, i.e. neural network algorithm meeting feedback error signal can be generated after input training sample, guarantee can be real-time
Correct weight.
The invention mainly comprises the following steps:
(1) first collect training connection data, unmanned surface vehicle is self-contained inertial navigation module, can be collected into the water surface nobody
2000 velocity informations before ship can be collected into 2000 times before unmanned surface vehicle wind speed and direction letters by the weather station of carrying
Breath, while the water flow meter by carrying can recorde the flow velocity of preceding 2000 water flows and flow to information, can also pass through carrying
Wave instrument obtain the information of wave height, wave direction and wavelength, can also be obtained by angular transducer propeller revolving speed letter
Breath.These information can be stored in designed program, pass through these available information of caller.
(2) it is grouped training using repetitive exercise mode, dividing six training datas first is one group, to BP nerve net
Network is trained, predict the 7th output data, then using nearest test sample together with next group of data as training data into
Row training, repetitive exercise method play the role of expanding training set, expand training set and are equivalent to the rule for expanding training data
Mould improves the accuracy rate of BP neural network.
(3) due to environmental disturbances etc., data set can inevitably have big number interference or noise error, therefore want
Hypothesis testing is carried out, the index of data set is identified, is handled, nondimensionalization.Nondimensionalization processing relatively common at present
Method mainly has extreme value, standardization, equalization and standard deviation.Standardized method is the most commonly used, but standardized method
Treated, and each index average is all 0, and standard deviation is all 1, it only reflects influencing each other between each index, in nondimensionalization
While also obliterated difference between each index in degree of variation.Therefore, using each finger of equalization method processing data set
Data are marked, the Eigen Covariance battle array R=cov (X) of primary standard sample set is calculated, seek the All Eigenvalues of covariance matrix R and right
The feature vector answered, and be ranked up according to characteristic value by substantially small.
(4) in collected training set data, since the variable of input is more, the standard of training result is often influenced whether
True property.Therefore using the Principal Component Analysis PCA in Multielement statistical analysis method, by multiple variables by linear transformation select compared with
Few number significant variable.Variance is obtained by the principal component of substantially small sequence, wherein the size of variance is characteristic value.To each spy
Value indicative carries out principal component analysis and casts out this feature value if weight E < 0.85.If weight E >=0.85, this feature value is chosen,
And calculate the variance accumulation serial number of this feature value.
(5) building and initialization of BP neural network establishes a trainable BP feedforward network with function, and function needs
Four input parameters.First parameter is the matrix of a R*2 to define the maximum value and minimum value of R input vector；Second
A parameter is the array of every layer of neuron number of a network；Third parameter is comprising every layer of transfer function title used
Cell array；The last one parameter is the title for the training function used, and Fig. 4 is neural metwork training state.
(6) repetitive exercise Jing Guo preceding 1800 groups of data, BP neural network have reached state that is stable and accurate and depositing,
The speed data within the scope of 1800-2000 group, the data result predicted are predicted at this time.
(7) data result of prediction is compared with real data, RMSE is asked to be used to measure observation between true value
Deviation root-mean-square error (standard error),Wherein x is average, and N is sample
Number.Inspection result is cast out the prediction result if RMSE < 60%, is grouped again to training set data, and neural network is strengthened
Training.If RMSE >=60% if prove that the prediction effect of the experiment reaches aspiration level, export the prediction knot of the training group
Fruit.
(8) judge whether grouping finishes.If being completed, experiment terminates.If not yet completing, returned data is grouped stage continuation
It is trained.
Claims (6)
1. a kind of unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network, it is characterized in that:
(1) data are collected, the parameter influential on predetermined speed of needs is picked out, specifically includes preceding 2000 groups of propeller
Revolving speed, the wind speed and direction of sea wind, the flow velocity of ocean current and flow direction, the wave height of wave and wavelength and wave direction four systems index；
(2) the four systems index is identified and is handled and the nondimensionalization of all indexs；
(3) principal component analysis is carried out to the system achievement data after four nondimensionalizations；
(4) unmanned surface vehicle speed of a ship or plane prediction BP neural network is initialized；
(5) network is trained with four systems index sample set；
(6) it tests to the generalization ability of unmanned surface vehicle speed of a ship or plane prediction BP neural network, is analyzed and corrected；
(7) BP neural network is predicted by the revised unmanned surface vehicle speed of a ship or plane, obtains the speed of subsequent time unmanned boat.
2. unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network according to claim 1, special
Sign is: being grouped training using repetitive exercise mode to the parameter influential on predetermined speed for picking out needs, draws first
Points of six training datas are one group, are trained to BP neural network, predict the 7th output data, then by nearest test specimens
Example is trained together with next group of data as training data.
3. unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network according to claim 1, special
Sign is described to be identified and handled and the nondimensionalization of all indexs specifically includes: handling data set using equalization method
Each achievement data, calculate primary standard sample set Eigen Covariance battle array R=cov (X), seek whole features of covariance matrix R
Value and corresponding feature vector, and be ranked up according to characteristic value by substantially small.
4. unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network according to claim 1, special
Sign is that the carry out principal component analysis specifically includes:, will be more using the Principal Component Analysis PCA in Multielement statistical analysis method
A variable selects significant variable by linear transformation, obtains variance by the principal component of substantially small sequence, wherein the size of variance is
It is characterized value, principal component analysis is carried out to each characteristic value and casts out this feature value if weight E < 0.85；If weight E >=
0.85, then this feature value is chosen, and calculate the variance accumulation serial number of this feature value.
5. unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network according to claim 1, special
Sign be it is described to the unmanned surface vehicle speed of a ship or plane prediction BP neural network carry out initialization specifically include: establishing one with function can instruct
Experienced BP feedforward network, function need four input parameters, first parameter be the matrix of a R*2 with define R it is a input to
The maximum value and minimum value of amount；Second parameter is the array of every layer of neuron number of a network；Third parameter is to include
The cell array of every layer of transfer function title used；The last one parameter is the title for the training function used.
6. unmanned surface vehicle speed of a ship or plane online forecasting method of the improvement based on BP neural network according to claim 1, special
Sign is that the generalization ability to unmanned surface vehicle speed of a ship or plane prediction BP neural network is tested, and is analyzed and is corrected tool
Body includes: that the data result of prediction compares with real data, and RMSE is asked to be used to measure observation with inclined between true value
The root-mean-square error of differenceWherein x is average, N is number of samples, inspection result, if
RMSE < 60% item casts out the prediction result, is grouped again to training set data, strengthens the training of neural network；If RMSE >
=60% proves that the prediction effect of the experiment reaches aspiration level, exports the prediction result of the training group.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810577099.XA CN108960421B (en) | 2018-06-05 | 2018-06-05 | Improved online forecasting method for speed of unmanned surface vehicle based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810577099.XA CN108960421B (en) | 2018-06-05 | 2018-06-05 | Improved online forecasting method for speed of unmanned surface vehicle based on BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108960421A true CN108960421A (en) | 2018-12-07 |
| CN108960421B CN108960421B (en) | 2022-03-18 |
Family
ID=64493892
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810577099.XA Active CN108960421B (en) | 2018-06-05 | 2018-06-05 | Improved online forecasting method for speed of unmanned surface vehicle based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108960421B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934422A (en) * | 2019-04-25 | 2019-06-25 | 中国人民解放军国防科技大学 | Neural network wind speed prediction method based on time series data analysis |
| CN110987066A (en) * | 2019-11-26 | 2020-04-10 | 青岛科技大学 | Ocean wind speed and direction measuring method and system capable of achieving automatic correction |
| CN111259943A (en) * | 2020-01-10 | 2020-06-09 | 天津大学 | Thermocline prediction method based on machine learning |
| CN111581832A (en) * | 2020-05-12 | 2020-08-25 | 哈尔滨工程大学 | Improved gray Elman neural network hovercraft motion prediction method based on ARMA model correction |
| CN113743662A (en) * | 2021-08-31 | 2021-12-03 | 中山大学 | Ship key technical parameter prediction method and system based on machine learning |
| CN113886973A (en) * | 2021-10-25 | 2022-01-04 | 江苏远望仪器集团有限公司 | Ship navigational speed processing method, device and processing equipment based on virtual-real mapping |
| CN117094219A (en) * | 2023-08-14 | 2023-11-21 | 中国科学院力学研究所 | Underwater cavity noise control method, device and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104200687A (en) * | 2014-09-11 | 2014-12-10 | 长安大学 | Driver speed control behavior monitoring device and method |
| KR20170087267A (en) * | 2016-01-20 | 2017-07-28 | 한국전자통신연구원 | Method for fast deep neural network computation using top n-best normalization |
| CN107481523A (en) * | 2017-09-27 | 2017-12-15 | 中南大学 | A kind of traffic flow speed Forecasting Methodology and system |
| CN107545250A (en) * | 2017-08-31 | 2018-01-05 | 哈尔滨工程大学 | A kind of ocean floating motion Real-time Forecasting System based on the remote sensing of wave image and artificial intelligence |
| CN107767946A (en) * | 2017-09-26 | 2018-03-06 | 浙江工业大学 | Breast cancer diagnosis system based on PCA (principal component analysis) and PSO-KE (particle swarm optimization-Key) L M (model-based regression) models |
| CN107944648A (en) * | 2018-01-08 | 2018-04-20 | 中国船舶工业系统工程研究院 | A kind of accurate Forecasting Methodology of large ship speed of a ship or plane rate of fuel consumption |
-
2018
- 2018-06-05 CN CN201810577099.XA patent/CN108960421B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104200687A (en) * | 2014-09-11 | 2014-12-10 | 长安大学 | Driver speed control behavior monitoring device and method |
| KR20170087267A (en) * | 2016-01-20 | 2017-07-28 | 한국전자통신연구원 | Method for fast deep neural network computation using top n-best normalization |
| CN107545250A (en) * | 2017-08-31 | 2018-01-05 | 哈尔滨工程大学 | A kind of ocean floating motion Real-time Forecasting System based on the remote sensing of wave image and artificial intelligence |
| CN107767946A (en) * | 2017-09-26 | 2018-03-06 | 浙江工业大学 | Breast cancer diagnosis system based on PCA (principal component analysis) and PSO-KE (particle swarm optimization-Key) L M (model-based regression) models |
| CN107481523A (en) * | 2017-09-27 | 2017-12-15 | 中南大学 | A kind of traffic flow speed Forecasting Methodology and system |
| CN107944648A (en) * | 2018-01-08 | 2018-04-20 | 中国船舶工业系统工程研究院 | A kind of accurate Forecasting Methodology of large ship speed of a ship or plane rate of fuel consumption |
Non-Patent Citations (2)
| Title |
|---|
| 王俊淑等: "高光谱遥感图像DE-self-training半监督分类算法", 《农业机械学报》 * |
| 钟洪亮: "大型船舶航速预报研究", 《中国优秀博硕士学位论文全文数据库 (硕士)工程科技Ⅱ辑》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934422A (en) * | 2019-04-25 | 2019-06-25 | 中国人民解放军国防科技大学 | Neural network wind speed prediction method based on time series data analysis |
| CN110987066A (en) * | 2019-11-26 | 2020-04-10 | 青岛科技大学 | Ocean wind speed and direction measuring method and system capable of achieving automatic correction |
| CN111259943A (en) * | 2020-01-10 | 2020-06-09 | 天津大学 | Thermocline prediction method based on machine learning |
| CN111581832A (en) * | 2020-05-12 | 2020-08-25 | 哈尔滨工程大学 | Improved gray Elman neural network hovercraft motion prediction method based on ARMA model correction |
| CN111581832B (en) * | 2020-05-12 | 2023-01-03 | 哈尔滨工程大学 | Improved gray Elman neural network hovercraft motion prediction method based on ARMA model correction |
| CN113743662A (en) * | 2021-08-31 | 2021-12-03 | 中山大学 | Ship key technical parameter prediction method and system based on machine learning |
| CN113743662B (en) * | 2021-08-31 | 2024-04-16 | 中山大学 | Ship key technical parameter prediction method and system based on machine learning |
| CN113886973A (en) * | 2021-10-25 | 2022-01-04 | 江苏远望仪器集团有限公司 | Ship navigational speed processing method, device and processing equipment based on virtual-real mapping |
| CN117094219A (en) * | 2023-08-14 | 2023-11-21 | 中国科学院力学研究所 | Underwater cavity noise control method, device and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108960421B (en) | 2022-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960421A (en) | The unmanned surface vehicle speed of a ship or plane online forecasting method based on BP neural network of improvement | |
| Khajavi et al. | Predicting the carbon dioxide emission caused by road transport using a Random Forest (RF) model combined by Meta-Heuristic Algorithms | |
| CN109214107A (en) | A kind of ship's navigation behavior on-line prediction method | |
| Yang et al. | AIS-based intelligent vessel trajectory prediction using bi-LSTM | |
| CN108563119B (en) | Unmanned ship motion control method based on fuzzy support vector machine algorithm | |
| CN118095972B (en) | Marine ecological environment prediction evaluation system based on artificial intelligence | |
| CN111695299B (en) | Mesoscale vortex trajectory prediction method | |
| CN106779137A (en) | A kind of method that ship oil consumption is predicted according to sea situation and operating condition | |
| Bakdi et al. | Fullest COLREGs evaluation using fuzzy logic for collaborative decision-making analysis of autonomous ships in complex situations | |
| CN111767517B (en) | BiGRU multi-step prediction method, system and storage medium applied to flood prediction | |
| CN109143408B (en) | Dynamic region combined short-time rainfall forecasting method based on MLP | |
| CN111310965A (en) | Aircraft track prediction method based on LSTM network | |
| CN113283653B (en) | Ship track prediction method based on machine learning and AIS data | |
| Suryo et al. | Improved time series prediction using LSTM neural network for smart agriculture application | |
| CN112766308A (en) | Ocean vessel classification method | |
| CN114152257A (en) | Ship prediction navigation method based on attention mechanism and environment perception LSTM | |
| CN113200086A (en) | Intelligent vehicle steering control system and control method thereof | |
| Vagale et al. | Evaluation of path planning algorithms of autonomous surface vehicles based on safety and collision risk assessment | |
| CN109977797A (en) | The optimization method of single order object detector based on sequence loss function | |
| Pérez et al. | Performance analysis of ANFIS in short term wind speed prediction | |
| CN117665735A (en) | Multi-source shipborne radar directional control method and device | |
| CN117275222A (en) | Traffic flow prediction method integrating one-dimensional convolution and attribute enhancement units | |
| Gunawan et al. | Long Short-Term Memory Approach for Predicting Air Temperature In Indonesia | |
| CN106342322B (en) | Based on the aircraft carrier fleet recognition methods of putting the reasoning of letter rule base | |
| CN109002860B (en) | Intelligent adaptive matching prediction method for sudden change wind speed along high-speed railway |
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 |