CN102289675A - Method for intelligently predicting ship course - Google Patents
Method for intelligently predicting ship course Download PDFInfo
- Publication number
- CN102289675A CN102289675A CN2011102072477A CN201110207247A CN102289675A CN 102289675 A CN102289675 A CN 102289675A CN 2011102072477 A CN2011102072477 A CN 2011102072477A CN 201110207247 A CN201110207247 A CN 201110207247A CN 102289675 A CN102289675 A CN 102289675A
- Authority
- CN
- China
- Prior art keywords
- vector machine
- course
- data
- square method
- yaw angle
- 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.)
- Pending
Links
Images
Abstract
The invention aims to provide a method for intelligently predicting a ship course. The method comprises the following steps of: acquiring ship yawing angle data, setting a sampling interval and setting a least squares support vector machine parameter; constructing a group of training data of a least squares support vector machine by taking yawing angles acquired at m moments before k-1 as inputs and a yawing angle psi(k) at a moment k as an output, correspondingly removing an earliest datum from the input once a new datum is acquired, and updating a model on time along with time to train the least squares support vector machine; applying a course predicting model of the trained least squares support vector machine to course prediction and taking a datum of a reach moment as an input to predict a ship yawing angle at a moment k+l+1. According to the method, the computing process is simplified greatly, and the realization of online real-time prediction is ensured.
Description
Technical field
What the present invention relates to is a kind of ship's navigation Forecasting Methodology.
Background technology
Boats and ships navigate by water environmental factors such as being subjected to wave, sea wind and disturb in wave, produce inevitably and wave, and athletic posture has very big randomness.Forecast the movement tendency on naval vessel in advance, significant to steady picture, carrier-borne aircraft landing task instruction and the compensation etc. of boats and ships imaging system.
Boats and ships are non-linear stochastic processes in marine motion, the motion model of being derived by ship's seakeeping theory and hydrodynamic methods is the approximate model under many assumed conditions, and to find the solution also be unusual difficulty, therefore uses such model can't provide the satisfied forecast of ship motion.Boats and ships exist a large amount of enchancement factors in the real navigation process, the ship course forecast is the time series modeling and the forecasting problem of a nonlinear system.Traditional time series analysis and prediction theory have effect preferably based on linear autoregression (AR) model and linear autoregressive moving average (ARMA) model to linear system, but are unsuitable for the time series modeling and the forecast of nonlinear system.Neural network also can be used for Nonlinear Modeling, but neural network mainly is based on the empiric risk minimization principle, theory shows that working as training data is tending towards for a long time infinite, empiric risk converges on practical risk, thus neural network impliedly used the infinite many assumed condition of sample based on the empiric risk minimization principle.In actual applications, this precondition often is not being met.
Support vector machine (Support Vector Machines, SVM) be a kind of new learning method that was applied to modeling in recent years, the optimum of SVM is found the solution based on structural risk minimization thought, therefore has than other nonlinear function approach methods and has stronger generalization ability.Finding the solution of SVM relates to quadratic programming problem, calculation of complex, efficient are low, least square method supporting vector machine (Least squares support vector machine, LSSVM) inequality constrain is converted into equality constraint, demand is separated a system of linear equations, the counting yield height realizes that ship course online forecasting LSSVM has more superiority than SVM.
Summary of the invention
A kind of ship course intelligent forecasting procedure that the object of the present invention is to provide accuracy height, computation process to simplify, can the online in real time forecast.
The object of the present invention is achieved like this:
The present invention 1, a kind of ship course intelligent forecasting procedure is characterized in that:
(1) gathers boats and ships yaw angle data, set sampling interval, and set the least square method supporting vector machine parameter;
(2) with m yaw angle constantly before the k-1 that gathers
As input, with k constantly yaw angle bow ψ (k) as exporting, one group of training data of structure least square method supporting vector machine, when collecting a new data, data are the earliest correspondingly removed from input, model carries out online updating accordingly as time passes, thus the training least square method supporting vector machine;
(3) the least square method supporting vector machine course forecasting model that trains is used for the course forecast, will
Arrive
Data constantly are as input, thereby predict k+l+1 boats and ships yaw angle constantly.
Advantage of the present invention is:
(1) because LSSVM has the ability of approaching any Nonlinear Mapping by study, modeling and identification with LSSVM is applied to nonlinear system can not be subjected to the restriction of nonlinear model.With the motion of LSSVM forecast non-linear stochastic process ship course, only need utilize the historical data of ship motion, Time Created, series model forecast the Ship Motion future value.
(2) carry out modeling according to the historical data of ship motion, forecast next course value constantly, along with the operation of system, old data constantly abandon, and new data constantly add, and use new data and set up the new model that can reflect system's the present situation, can guarantee the accuracy of model.
(3) can well to solve neural metwork training speed at the small sample training slow for least square method supporting vector machine, be absorbed in shortcomings such as local extremum easily, and the solving-optimizing problem finally transfers to finds the solution linear equation, computation process has obtained great simplification, has guaranteed to realize online real-time prediction.
Description of drawings
Fig. 1 is a process flow diagram of the present invention.
Fig. 2 (a) is 3.8 meters of adopted wave height for wave has, the comparison diagram of BP neural network yaw angle predicted value and actual value when experience wave-to-course angle 30 is spent, Fig. 2 (b) is 3.8 meters of adopted wave height for wave has, the comparison diagram of BP neural network yaw angle predicted value and actual value when experience wave-to-course angle 90 is spent, Fig. 2 (c) is 3.8 meters of adopted wave height for wave has, the comparison diagram of BP neural network yaw angle predicted value and actual value when meeting with wave-to-course angle 150 and spending;
Fig. 3 (a) is 3.8 meters of adopted wave height for wave has, the comparison diagram of LSSVM yaw angle predicted value and actual value when experience wave-to-course angle 30 is spent, Fig. 3 (b) is 3.8 meters of adopted wave height for wave has, the comparison diagram of LSSVM yaw angle predicted value and actual value when experience wave-to-course angle 90 is spent, Fig. 3 (c) is 3.8 meters of adopted wave height for wave has, the comparison diagram of LSSVM yaw angle predicted value and actual value when meeting with wave-to-course angle 150 and spending.
Embodiment
For example the present invention is done description in more detail below in conjunction with accompanying drawing:
In conjunction with Fig. 1～3, the present invention includes following steps:
(1) gather boats and ships yaw angle data, sampling interval is 0.1 second.Set the least square method supporting vector machine parameter: kernel function parameter σ and penalty factor γ.
(2) with m yaw angle constantly before the k-1 that gathers
As input, with k constantly yaw angle bow ψ (k) as exporting, one group of training data of structure least square method supporting vector machine, when collecting a new data, data are the earliest correspondingly removed from input, along with the operation of system, training data constantly upgrades, model also carries out online updating accordingly as time passes, utilizes the l group training data of structure, the training least square method supporting vector machine.
(3) the least square method supporting vector machine course forecasting model that trains is used for the course forecast, will
Arrive
Data are constantly predicted k+l+1 boats and ships yaw angle constantly as input.
1. least square method supporting vector machine regression algorithm
For training data
YP R, n are number of samples.The least square method supporting vector machine Nonlinear Mapping
With the feature space of sample, and in the feature space of this higher-dimension, be constructed as follows the linear regression function and realize to the match of sample data and to the prediction of to-be from former spatial mappings to higher-dimension.
In the formula:
Be nonlinear function, input is mapped to feature space; ω, b represent weight coefficient and biasing respectively.
LSSVM constructs the regression function of formula (1) by finding the solution following constrained optimization problem.
In the formula: ξ
kIt is relaxation factor; γ is the punishment parameter, is implemented in the regression error of permission and the compromise between the algorithm complex.
Definition Lagrange function:
(3)
In the formula:
It is the Lagrange operator.Following formula is optimized, can be write as through conversion:
Wherein,
Note
Element in the matrix power has following form:
Through finding the solution of above-mentioned system of equations, can obtain following LSSVM regression model at last:
Its α, b is solved by formula (4), and y (x) is based on system's output of LSSVM model, α in the training sample
kNon-vanishing sample is exactly a support vector, kernel function K (x, x
k) purpose is to extract feature from luv space, the sample in the luv space is mapped as a vector in the high-dimensional feature space, to solve the inseparable problem of luv space neutral line.
2. ship course LSSVM online forecasting model
For the ship course exercise data
Be the forecast desired value, set up input
And output
Between mapping relations.
In the LSSVM forecasting model, the sample of least square method supporting vector machine study is:
Wherein, m is the input dimension, and n is the training sample number.
After training is finished, be to going on foot forecast one of future:
The forecast of second step is:
By that analogy, the forecast model in p step is:
In the formula, x
nThe actual value of representing n data,
The predicted value of representing n data.
Embodiment:
(1) gather boats and ships yaw angle data, sampling interval is 0.1 second, and simulation time is 100 seconds, therefore can obtain 1000 yaw angle data.Utilize preceding 500 groups as training data, the back 500 groups as test data.Set the least square method supporting vector machine parameter: kernel function parameter σ and penalty factor γ.
(2) utilize least square method supporting vector machine forecast ship course, with the input of yaw angle ψ (k) as LSSVM, input vector is:
Wherein, m is the input dimension.
Least square method supporting vector machine is output as
Given training sample:
Wherein, X
1,,, X
lBe that k arrives
Input vector constantly:
Utilize the training data of structure, the training least square method supporting vector machine.
(3) the least square method supporting vector machine course forecasting model that trains is used for the course forecast, will
Arrive
Data are constantly predicted k+l+1 boats and ships yaw angle constantly as input.
Fig. 1 is least square method supporting vector machine ship course forecasting procedure figure.
(a) and (b), (c) are respectively that wave has 3.8 meters of adopted wave height among Fig. 2, Fig. 3, meet with the comparison diagram of using BP neural network and LSSVM forecast boats and ships yaw angle predicted value and actual value when wave-to-course angle 30,90,150 is spent.Solid line is an actual value among the figure, and dotted line is a predicted value.The X-axis express time, unit 0.1 second; Y-axis is represented yaw angle, and unit is degree.
Can find out in conjunction with Fig. 2, Fig. 3, the forecast result who adopts the LSSVM method obviously since the BP neural network the forecast result, can effectively improve forecast precision based on the ship course forecasting procedure of least square method supporting vector machine.
Claims (1)
1. ship course intelligent forecasting procedure is characterized in that:
(1) gathers boats and ships yaw angle data, set sampling interval, and set the least square method supporting vector machine parameter;
(2) with m yaw angle ψ (k-1) constantly before the k-1 that gathers, ψ (k-m-1) is as input, with k constantly yaw angle bow ψ (k) as exporting, one group of training data of structure least square method supporting vector machine, when collecting a new data, data are the earliest correspondingly removed from input, model carries out online updating accordingly as time passes, thus the training least square method supporting vector machine;
(3) the least square method supporting vector machine course forecasting model that trains is used for the course forecast, k+l is arrived k-m+l data constantly as input, thereby predict k+l+1 boats and ships yaw angle constantly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011102072477A CN102289675A (en) | 2011-07-24 | 2011-07-24 | Method for intelligently predicting ship course |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011102072477A CN102289675A (en) | 2011-07-24 | 2011-07-24 | Method for intelligently predicting ship course |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102289675A true CN102289675A (en) | 2011-12-21 |
Family
ID=45336081
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011102072477A Pending CN102289675A (en) | 2011-07-24 | 2011-07-24 | Method for intelligently predicting ship course |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102289675A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103322415A (en) * | 2013-06-05 | 2013-09-25 | 哈尔滨工程大学 | Two-dimensional reproduction method for petroleum pipeline defects through least squares support vector machines (LS-SVM) |
| CN103926932A (en) * | 2014-04-25 | 2014-07-16 | 哈尔滨工程大学 | Intelligent ship moving posture decomposition field forecasting method |
| CN104049639A (en) * | 2014-06-24 | 2014-09-17 | 上海大学 | Unmanned surface vehicle anti-surge control device and method based on support vector regression |
| CN107145690A (en) * | 2017-06-12 | 2017-09-08 | 郑州云海信息技术有限公司 | A kind of control method of carrier-based helicopter from motion tracking Ship Motion |
| CN108802040A (en) * | 2017-05-04 | 2018-11-13 | 南京市特种设备安全监督检验研究院 | A kind of unmanned plane device and detection method for crane surface defects detection |
| CN113433934A (en) * | 2021-04-27 | 2021-09-24 | 武汉海兰鲸科技有限公司 | Method for optimizing navigational speed of commercial ship |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7020642B2 (en) * | 2002-01-18 | 2006-03-28 | Pavilion Technologies, Inc. | System and method for pre-processing input data to a support vector machine |
| CN101859105A (en) * | 2010-06-21 | 2010-10-13 | 哈尔滨工程大学 | On-line forecasting method of fault of ship course control system |
| CN101859147A (en) * | 2010-06-11 | 2010-10-13 | 哈尔滨工程大学 | Ship course intelligent coordination control method |
| CN102065449A (en) * | 2010-12-13 | 2011-05-18 | 哈尔滨工业大学 | Method for predicting mobile communication telephone traffic based on clustered LS-SVM (Least Squares-Support Vector Machine) |
-
2011
- 2011-07-24 CN CN2011102072477A patent/CN102289675A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7020642B2 (en) * | 2002-01-18 | 2006-03-28 | Pavilion Technologies, Inc. | System and method for pre-processing input data to a support vector machine |
| CN101859147A (en) * | 2010-06-11 | 2010-10-13 | 哈尔滨工程大学 | Ship course intelligent coordination control method |
| CN101859105A (en) * | 2010-06-21 | 2010-10-13 | 哈尔滨工程大学 | On-line forecasting method of fault of ship course control system |
| CN102065449A (en) * | 2010-12-13 | 2011-05-18 | 哈尔滨工业大学 | Method for predicting mobile communication telephone traffic based on clustered LS-SVM (Least Squares-Support Vector Machine) |
Non-Patent Citations (1)
| Title |
|---|
| 杜颖等: "基于最小二乘支持向量机的风电场短期风速预测", 《电网技术》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103322415A (en) * | 2013-06-05 | 2013-09-25 | 哈尔滨工程大学 | Two-dimensional reproduction method for petroleum pipeline defects through least squares support vector machines (LS-SVM) |
| CN103926932A (en) * | 2014-04-25 | 2014-07-16 | 哈尔滨工程大学 | Intelligent ship moving posture decomposition field forecasting method |
| CN104049639A (en) * | 2014-06-24 | 2014-09-17 | 上海大学 | Unmanned surface vehicle anti-surge control device and method based on support vector regression |
| CN104049639B (en) * | 2014-06-24 | 2016-12-07 | 上海大学 | A kind of unmanned boat antisurge based on support vector regression controls apparatus and method |
| CN108802040A (en) * | 2017-05-04 | 2018-11-13 | 南京市特种设备安全监督检验研究院 | A kind of unmanned plane device and detection method for crane surface defects detection |
| CN107145690A (en) * | 2017-06-12 | 2017-09-08 | 郑州云海信息技术有限公司 | A kind of control method of carrier-based helicopter from motion tracking Ship Motion |
| CN107145690B (en) * | 2017-06-12 | 2020-10-20 | 苏州浪潮智能科技有限公司 | Control method for automatically tracking movement of ship-based helicopter |
| CN113433934A (en) * | 2021-04-27 | 2021-09-24 | 武汉海兰鲸科技有限公司 | Method for optimizing navigational speed of commercial ship |
| CN113433934B (en) * | 2021-04-27 | 2022-04-12 | 武汉海兰鲸科技有限公司 | Method for optimizing navigational speed of commercial ship |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102289675A (en) | Method for intelligently predicting ship course | |
| CN102411305B (en) | Design method of comprehensive disturbance rejection control system for single-rotor wing helicopter/turboshaft engine | |
| Li et al. | Ship motion prediction using dynamic seasonal RvSVR with phase space reconstruction and the chaos adaptive efficient FOA | |
| CN107092987B (en) | Method for predicting autonomous landing wind speed of small and medium-sized unmanned aerial vehicles | |
| CN104049639B (en) | A kind of unmanned boat antisurge based on support vector regression controls apparatus and method | |
| CN108197350A (en) | A kind of unmanned boat speed and uncertainty estimation system and design method | |
| CN112182972B (en) | Ship maneuvering motion ADAM local weighted regression identification modeling method | |
| Cheng et al. | A neural-network-based sensitivity analysis approach for data-driven modeling of ship motion | |
| CN104865825B (en) | A kind of multi-Robot Cooperative time series forecasting control method | |
| CN115359338A (en) | Sea surface temperature prediction method and system based on hybrid learning model | |
| CN109144099A (en) | Unmanned aerial vehicle group action scheme fast evaluation method based on convolutional neural networks | |
| Liu et al. | A prediction method for deck-motion based on online least square support vector machine and genetic algorithm | |
| CN106803361A (en) | A kind of navigation method of control based on Rolling Planning strategy | |
| Jeong et al. | A shipyard simulation system using the process-centric simulation modeling methodology: Case study of the simulation model for the shipyard master plan validation | |
| Zhang et al. | Real-time simulation of a rescue ship maneuvering in short-crested irregular waves | |
| CN112099506A (en) | Tracking control method and system for under-actuated unmanned ship time-varying formation | |
| Huang et al. | Short-term prediction of ship pitching motion based on artificial neural networks | |
| CN113836661A (en) | Time prediction method, model training method, related device and electronic equipment | |
| Junji et al. | Accurate calculating method of marine three-component geomagnetic field in shipboard measurement | |
| Liu et al. | Improving deterministic pitch motions estimation using bivariate sequential wave input | |
| Guo et al. | Forward kinematics analysis of the stewart parallel platform based on the elman recurrent network | |
| Jun-jie | The optimization DGM (2, 1) model and its application | |
| Wang et al. | The prediction method of tropical cyclone intensity change based on deep learning | |
| Liu et al. | Short-term prediction of ship motion based on EMD-SVM | |
| Huang et al. | A hybrid grey relational analysis and support vector machines approach for forecasting consumption of spare parts |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20111221 |