CN107065569B - Ship dynamic positioning sliding mode control system and method based on RBF neural network compensation - Google Patents
Ship dynamic positioning sliding mode control system and method based on RBF neural network compensation Download PDFInfo
- Publication number
- CN107065569B CN107065569B CN201710396708.7A CN201710396708A CN107065569B CN 107065569 B CN107065569 B CN 107065569B CN 201710396708 A CN201710396708 A CN 201710396708A CN 107065569 B CN107065569 B CN 107065569B
- Authority
- CN
- China
- Prior art keywords
- ship
- sliding mode
- neural network
- control
- dynamic positioning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Abstract
The invention provides a ship dynamic positioning sliding mode control system and method based on RBF neural network compensation. The dynamic positioning system comprises a display control computer (1), a guide system (2), an extended state observer (3), a sliding mode controller (4), an RBF neural network compensator (5), an executing mechanism (6), a dynamic positioning ship (7) and a sensor system (8). The ship dynamic positioning sliding mode control system and method based on RBF neural network compensation consider the problems of unmodeled dynamics, uncertain models, environmental interference and saturated input of an actuating mechanism in ship motion, improve the anti-interference capability and positioning accuracy of the ship dynamic positioning system, and can achieve better control effect under the condition that the input of actuating mechanisms such as a propeller and a propeller is limited.
Description
Technical Field
The invention relates to a ship dynamic positioning sliding mode control system and a ship dynamic positioning sliding mode control method.
Background
The exploitation of marine resources such as offshore oil, natural gas and the like opens a new era for the development of ship dynamic positioning, the number of ships equipped with the dynamic positioning control system is from the first few to the present thousands or even more, and the wide application prospect of the dynamic positioning control system is fully displayed. The dynamic positioning control system is widely applied to operations such as offshore oil development, core drilling, submarine mining, cable laying, pipe laying, diving support, offshore fire fighting and the like. The dynamic positioning system is used for keeping the position and the heading of a ship only through the thrust generated by a propeller or a propeller of the dynamic positioning system. In the dynamic positioning motion control of the ship, the control effect of the ship motion is influenced by model uncertainty caused by hydrodynamic parameters and other factors, unmodeled dynamic terms in a mathematical model, slowly-varying environmental interference and the like, and the input of a control system is limited due to the amplitude limiting problem of actuating mechanisms such as a propeller and the like, so that the research on the problems has certain practical significance.
Chinese patent CN103760900A proposes a ship motion control system considering control input constraint, which utilizes the combination of a filtering backstepping method and a self-adaptive neural network to control the ship motion, and solves the input constraint problem and the unknown nonlinear problem of a ship model in the control process. Different from the method, the method mainly aims at the problems of uncertain models, unmodeled dynamic terms and environmental interference in the ship motion under the condition of saturated execution mechanisms.
Chinese patent CN105867382A proposes an equivalent interference compensation based ship dynamic positioning control system, which equates external environment interference, internal system uncertainty items and interference items into a state variable to respectively design extended state observers for three degrees of freedom of ship motion, thereby improving the anti-interference capability of the control system. Different from the method, the method not only researches uncertain models, unmodeled dynamic terms and environmental interference problems in ship motion, but also researches saturation problems caused by actuating mechanisms such as a propeller and a propeller in a control system.
Disclosure of Invention
The invention aims to provide a ship dynamic positioning sliding mode control system based on RBF neural network compensation, which has good anti-interference capability and high positioning precision and can achieve better control effect under the condition of limited input of an actuating mechanism. The invention also aims to provide a ship dynamic positioning sliding mode control method based on RBF neural network compensation.
The ship dynamic positioning sliding mode control system based on RBF neural network compensation comprises a display control computer 1, a guide system 2, an extended state observer 3, a sliding mode controller 4, an RBF neural network compensator 5, an execution mechanism 6, a dynamic positioning ship 7 and a sensor system 8; the method is characterized in that: the sensor system 8 collects position and angle information of the dynamic positioning ship 7 in real time, the position and angle information is called pose information eta for short, and the collected pose information is transmitted to the display control computer 1 and the extended state observer 3; the display and control computer 1 displays the actual pose signal of the ship in real time and sends an expected pose step signal etad0To the guidance system 2; the guidance system 2 carries out smoothing processing on the step signal of the expected pose to obtain continuous expected pose information etadAnd first and second derivatives thereofAnd transmitted to the sliding mode controller 4; the extended state observer 3 expands unmodeled dynamics, model uncertainty and environmental interference in ship motion into an extended state vector d, and estimates attitude information and the extended state vector to obtain the extended state vectorState vector estimationPose information estimation value and first-order and second-order derivatives thereofAnd transmitted to the sliding mode controller 4; the RBF neural network compensator 5 performs RBF approximation on the control quantity error under the control input saturation to obtain a control quantity compensation errorAnd transmitted to the sliding mode controller 4; the sliding mode controller 4 aims at the expected pose information and the first and second derivatives eta thereofd,Pose information estimation value and first-order and second-order derivative and expansion state vector estimation value thereofControl quantity error estimation under control input saturationPerforming sliding mode control to obtain a control quantity tau of the actuating mechanism 6 under an input saturation condition; the executing mechanism 6 controls the dynamic positioning ship 7 according to the control quantity tau output by the sliding mode controller 4, so that the ship moves to a desired pose state.
The ship dynamic positioning sliding mode control method based on RBF neural network compensation comprises the following steps:
(1) the sensor system 8 collects position and angle information of the dynamic positioning ship 7 in real time, the position and angle information is called pose information eta for short, and the collected pose information is transmitted to the display control computer 1 and the extended state observer 3;
(2) the display and control computer 1 displays the actual pose signal of the ship in real time and sends an expected pose step signal etad0To the guidance system 2;
(3) the guidance system 2 carries out smoothing processing on the step signal of the expected poseObtaining continuous expected pose information etadAnd first and second derivatives thereofAnd transmitted to the sliding mode controller 4;
(4) the extended state observer 3 expands unmodeled dynamics, model uncertainty items and environmental interference in ship motion into an extended state vector d, and estimates attitude information and the extended state vector to obtain an extended state vector estimation valuePose information estimation value and first-order and second-order derivatives thereofAnd transmitted to the sliding mode controller 4;
(5) the RBF neural network compensator 5 performs RBF approximation on the control quantity error under the control input saturation to obtain a control quantity compensation errorAnd transmitted to the sliding mode controller 4;
(6) the sliding mode controller 4 aims at the expected pose information and the first and second derivatives eta thereofd,Pose information estimation value and first-order and second-order derivative and expansion state vector estimation value thereofControl quantity error estimation under control input saturationPerforming sliding mode control to obtain a control quantity tau of the actuating mechanism 6 under an input saturation condition;
(7) the executing mechanism 6 controls the dynamic positioning ship 7 according to the control quantity tau output by the sliding mode controller 4, so that the ship moves to a desired pose state.
The invention can improve the anti-interference capability and the positioning precision of the ship dynamic positioning system, and can achieve better control effect under the condition that the input of actuating mechanisms such as a propeller and a propeller is limited.
Drawings
FIG. 1 is a general structure diagram of a ship dynamic positioning sliding mode control system based on RBF neural network compensation;
FIG. 2 is a motion profile of a dynamically positioned vessel at saturation without regard to control inputs;
FIG. 3 is a motion profile of a dynamically positioned vessel under consideration of control input saturation;
FIG. 4 is a dynamic positioning vessel longitudinal control force curve;
FIG. 5 is a dynamic positioning vessel lateral control force curve;
FIG. 6 is a dynamic positioning vessel heading control moment curve.
Detailed Description
The invention is described in more detail below with reference to the accompanying drawings.
The ship dynamic positioning sliding mode control system based on RBF neural network compensation comprises a display control computer 1, a guide system 2, an extended state observer 3, a sliding mode controller 4, an RBF neural network compensator 5, an execution mechanism 6, a dynamic positioning ship 7 and a sensor system 8. The sensor system 8 collects position and angle information of the dynamic positioning ship 7 in real time, the position and angle information is called pose information eta for short, and the collected pose information is transmitted to the display control computer 1 and the extended state observer 3; the display and control computer 1 displays the actual pose signal of the ship in real time and sends an expected pose step signal etad0To the guidance system 2; the guidance system 2 carries out smoothing processing on the step signal of the expected pose to obtain continuous expected pose information etadAnd first and second derivatives thereofAnd transmitted to the sliding mode controller 4; the extended state observer 3 expands unmodeled dynamics, model uncertainty and environmental disturbance in the ship motion into an extended state vector d, and aligns attitude information and extensionEstimating the tensor state vector to obtain an estimated value of the expansion state vectorPose information estimation value and first-order and second-order derivatives thereofAnd transmitted to the sliding mode controller 4; the RBF neural network compensator 5 performs RBF approximation on the control quantity error under the control input saturation to obtain a control quantity compensation errorAnd transmitted to the sliding mode controller 4; the sliding mode controller 4 aims at the expected pose information and the first and second derivatives eta thereofd,Pose information estimation value and first-order and second-order derivative and expansion state vector estimation value thereofControl quantity error estimation under control input saturationPerforming sliding mode control to obtain a control quantity tau of the actuating mechanism 6 under an input saturation condition; the executing mechanism 6 controls the dynamic positioning ship 7 according to the control quantity tau output by the sliding mode controller 4, so that the ship moves to a desired pose state.
The ship dynamic positioning sliding mode control method based on RBF neural network compensation of the invention is explained in more detail as follows:
the ship three-degree-of-freedom (surging, swaying and yawing) low-frequency motion model is as follows:
wherein the content of the first and second substances,the position and attitude vectors of the ship under an inertial coordinate system (x, y are the transverse and longitudinal positions of the ship, psi is the ship heading angle),is the first derivative of the attitude information of the ship in the inertial system,is the pose vector of the ship under the attached coordinate system (u, v are the transverse and longitudinal speeds of the ship, r is the heading angular speed of the ship),in the form of a ship mass matrix,D(υ)>03×3,D(υ)=DΤ(v) is a matrix of damping coefficients,is a matrix of coriolis and center forces,in order to control the force and moment vectors,force and moment vectors generated by slowly varying environmental disturbances and unmodeled dynamics.Is a transformation matrix between the earth-fixed coordinate system and the satellite coordinate system, and the concrete representation form is
Step signal eta of guidance system for outputting expected pose to display control computerd0Smoothing to obtain expected pose informationAnd first and second derivatives of expected pose information
Wherein, TsAs cutoff time, ωnAnd zeta is the relative damping ratio of the oscillation link of the guide system and the design parameter of the longitudinal path generator.
In order to facilitate the control of the ship, the ship model (1) is transformed as follows:
wherein the content of the first and second substances,for transforming the derivative of the matrix R (ψ) with respect to time, and
M*=R(ψ)MRΤ(ψ)
D*=R(ψ)D(υ)RΤ(ψ)
b*=R(ψ)RΤ(ψ)b=b
the specific form of the state observer is as follows:
wherein the content of the first and second substances,is an estimated value of the position and orientation vector of the ship,is an estimated value of a first-order derivative of the ship position and attitude signal,is the first derivative of the estimated value of the ship position vector,is the second derivative of the estimated value of the ship position vector,in order to expand the estimated value of the state variable,the first derivative of the estimate of the dilated state variable,is a positive definite matrix and > 0 is a constant.
Setting the error between the actual control input and the control quantity asWherein sat (tau) is an input saturation function in the specific form
The RBF neural network compensator approximates the error between the control quantities to obtain a compensation valueThe RBF control algorithm is
Wherein the content of the first and second substances,for neural network input, hjJ is 1,2, …, l is the output vector of the gaussian vector basis function, cjAs the center position of the jth hidden layer neuron, bjThe width of Gaussian distribution is less than or equal tomaxFor neural network error estimation, the weight matrix of the neural network is
Wherein the content of the first and second substances,for the weight estimation matrix, h (x) is a Gaussian basis vector.
Wherein the content of the first and second substances,υris the intermediate variable(s) of the variable,is a gain matrix.
Therefore, the sliding mode control law is
Wherein, is a constant value square matrix, sΤIs the transpose of the sliding mode function vector.
Claims (2)
1. A ship dynamic positioning sliding-mode control system based on RBF neural network compensation comprises a display control computer (1), a guide system (2), an extended state observer (3), a sliding-mode controller (4), an RBF neural network compensator (5), an execution mechanism (6), a dynamic positioning ship (7) and a sensor system (8); the method is characterized in that: the sensor system (8) collects position and angle information of the dynamic positioning ship (7) in real time, the position and angle information is called pose information eta for short, and the collected pose information is transmitted to the display control computer (1) and the extended state observer (3); the display control computer (1) displays the actual pose signal of the ship in real time and sends an expected pose step signal etad0To the guidance system (2); the guidance system (2) carries out step signal processing on the expected posePerforming smoothing treatment to obtain continuous expected pose information etadAnd first and second derivatives thereofAnd transmitted to the sliding mode controller (4); the extended state observer (3) expands unmodeled dynamics, model uncertainty and environmental interference in ship motion into an extended state vector d, and estimates attitude information and the extended state vector to obtain an extended state vector estimation valuePose information estimation value and first-order and second-order derivatives thereofAnd transmitted to the sliding mode controller (4); the RBF neural network compensator (5) carries out RBF approximation aiming at the control quantity error under the control input saturation to obtain the control quantity compensation errorAnd transmitted to the sliding mode controller (4); the sliding mode controller (4) aims at the expected pose information and the first-order and second-order derivatives eta thereofd,Pose information estimation value and first-order and second-order derivative and expansion state vector estimation value thereofControl quantity error estimation under control input saturationPerforming sliding mode control to obtain a control quantity tau of the actuating mechanism (6) under an input saturation condition; the executing mechanism (6) controls the dynamic positioning ship (7) according to the control quantity tau output by the sliding mode controller (4) so that the ship moves to a desired pose state;
the ship three-degree-of-freedom low-frequency motion model is as follows:
wherein the content of the first and second substances,is the position and attitude vector of the ship under an inertial coordinate system, x and y are the transverse and longitudinal positions of the ship, psi is the ship heading angle,is the first derivative of the attitude information of the ship in the inertial system,is the pose vector of the ship in the attached coordinate system, u and v are the transverse and longitudinal speeds of the ship, r is the heading angular speed of the ship,in the form of a ship mass matrix,D(υ)>03×3,D(υ)=DΤ(v) is a matrix of damping coefficients,is a matrix of coriolis and center forces,in order to control the force and moment vectors,for the force and moment vectors generated by slowly varying environmental disturbances and unmodeled dynamics,the transformation matrix is a transformation matrix between a ground-fixed coordinate system and a random coordinate system, and the concrete representation form is as follows:
wherein, TsAs cutoff time, ωnThe natural frequency of the oscillation link of the guide system, zeta is the relative damping ratio of the oscillation link of the guide system, and is a design parameter of the longitudinal path generator;
the specific form of the state observer is as follows:
wherein the content of the first and second substances,is an estimated value of the position and orientation vector of the ship,is an estimated value of a first-order derivative of the ship position and attitude signal,is the first derivative of the estimated value of the ship position vector,is the second derivative of the estimated value of the ship position vector,in order to expand the estimated value of the state variable,the first derivative of the estimate of the dilated state variable,i is 1,2 and 3 are positive definite matrixes, and more than 0 is a constant;
the RBF control algorithm is
={Wi *}Τ·{hi(x)}+
Wherein the content of the first and second substances,for neural network input, hjJ is 1,2, …, l is the output vector of the gaussian vector basis function, cjAs the center position of the jth hidden layer neuron, bjThe width of Gaussian distribution is less than or equal tomaxFor neural network error estimation, the weight matrix of the neural network is
the sliding mode control law is as follows:
Defining:
D*=R(ψ)D(υ)RΤ(ψ)
b*=R(ψ)RΤ(ψ)b=b
for a transformation matrix, R, between the earth-fixed coordinate system and the satellite coordinate systemΤ(psi) is a transposed matrix of R (psi),is the first derivative of R (ψ) with respect to time,m is a ship mass matrix,D(υ)>03×3,D(υ)=DΤ(upsilon), D (upsilon) is a damping coefficient matrix,c (upsilon) is the coriolis and center force matrix,force and moment vectors generated by slowly varying environmental disturbances and unmodeled dynamics;
υris the intermediate variable(s) of the variable,for an intermediate variable, letηdDefining errors for expected pose information As an estimated value of the pose information,is a gain matrix, thenIs a medium variable upsilonrA first derivative with respect to time;
wherein, is a constant value square matrix, sΤIs the transpose of the sliding mode function vector.
2. The control method of the ship dynamic positioning sliding-mode control system based on the RBF neural network compensation as claimed in claim 1, is characterized in that:
(1) the sensor system (8) collects position and angle information of the dynamic positioning ship (7) in real time, the position and angle information is called pose information eta for short, and the collected pose information is transmitted to the display control computer (1) and the extended state observer (3);
(2) the display control computer (1) displays the actual pose signal of the ship in real time and sends an expected pose step signal etad0To the guidance system (2);
(3) the guidance system (2) carries out smoothing processing on the step signal of the expected pose to obtain continuous expected pose information etadAnd first and second derivatives thereofAnd transmitted to the sliding mode controller (4);
(4) the extended state observer (3) expands unmodeled dynamics, model uncertainty and environmental interference in ship motion into an extended state vector d, and estimates attitude information and the extended state vector to obtain an extended state vector estimation valuePose information estimation value and first-order and second-order derivatives thereofAnd transmitted to the sliding mode controller (4);
(5) the RBF neural network compensator (5) is used for controlling the error of the control quantity under the saturation of the control inputRBF approximation is carried out to obtain the compensation error of the control quantityAnd transmitted to the sliding mode controller (4);
(6) the sliding mode controller (4) aims at the expected pose information and the first-order and second-order derivatives eta thereofd,Pose information estimation value and first-order and second-order derivative and expansion state vector estimation value thereofControl quantity error estimation under control input saturationPerforming sliding mode control to obtain a control quantity tau of the actuating mechanism (6) under an input saturation condition;
(7) and the executing mechanism (6) controls the dynamic positioning ship (7) according to the control quantity tau output by the sliding mode controller (4), so that the ship moves to a desired pose state.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710396708.7A CN107065569B (en) | 2017-05-31 | 2017-05-31 | Ship dynamic positioning sliding mode control system and method based on RBF neural network compensation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710396708.7A CN107065569B (en) | 2017-05-31 | 2017-05-31 | Ship dynamic positioning sliding mode control system and method based on RBF neural network compensation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107065569A CN107065569A (en) | 2017-08-18 |
CN107065569B true CN107065569B (en) | 2020-11-03 |
Family
ID=59616734
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710396708.7A Active CN107065569B (en) | 2017-05-31 | 2017-05-31 | Ship dynamic positioning sliding mode control system and method based on RBF neural network compensation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107065569B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108594639B (en) * | 2018-03-28 | 2020-12-22 | 哈尔滨工程大学 | Reinforced learning-based track tracking control method for hovercraft |
CN109100939B (en) * | 2018-09-19 | 2021-05-11 | 哈尔滨工程大学 | Input saturation considered water surface unmanned ship all-state constraint trajectory tracking control method |
CN109828463A (en) * | 2019-02-18 | 2019-05-31 | 哈尔滨工程大学 | A kind of adaptive wave glider bow of ocean current interference is to control method |
CN110032075B (en) * | 2019-05-28 | 2022-04-05 | 哈尔滨工程大学 | Design method of saturation compensation control system of dynamic positioning ship |
CN111381595B (en) * | 2020-03-10 | 2023-07-25 | 大连海事大学 | Ship dynamic positioning method based on event triggering |
CN111736617B (en) * | 2020-06-09 | 2022-11-04 | 哈尔滨工程大学 | Track tracking control method for preset performance of benthonic underwater robot based on speed observer |
CN113406884B (en) * | 2021-06-03 | 2023-03-24 | 上海海事大学 | Sliding mode self-adaptation-based multi-point mooring system positioning control method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105867382B (en) * | 2016-05-12 | 2018-08-31 | 哈尔滨工程大学 | A kind of ship power-positioning control system based on equivalent interference compensation |
-
2017
- 2017-05-31 CN CN201710396708.7A patent/CN107065569B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN107065569A (en) | 2017-08-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107065569B (en) | Ship dynamic positioning sliding mode control system and method based on RBF neural network compensation | |
Du et al. | Robust dynamic positioning of ships with disturbances under input saturation | |
Do et al. | Underactuated ships follow smooth paths with integral actions and without velocity measurements for feedback: theory and experiments | |
CN109917657B (en) | Anti-interference control method and device for dynamic positioning ship and electronic equipment | |
Hu et al. | Robust adaptive NN control of dynamically positioned vessels under input constraints | |
CN110032075B (en) | Design method of saturation compensation control system of dynamic positioning ship | |
Fortuna et al. | A roll stabilization system for a monohull ship: modeling, identification, and adaptive control | |
CN106054884B (en) | L1 self-adaptive ship dynamic positioning Double Loop Control System neural network based | |
CN109343347B (en) | Track tracking control method for submarine flight nodes | |
Kyrkjebø et al. | Output synchronization control of ship replenishment operations: Theory and experiments | |
CN110865539A (en) | Unmanned ship tracking error constraint control method under random interference | |
CN109739248B (en) | Ship-borne three-degree-of-freedom parallel stable platform stability control method based on ADRC | |
Nguyen et al. | Switching control for thruster-assisted position mooring | |
Ngongi et al. | Linear fuzzy controller design for dynamic positioning system of surface ships | |
Liu et al. | A hierarchical disturbance rejection depth tracking control of underactuated AUV with experimental verification | |
Zhang et al. | Tracking fault-tolerant control based on model predictive control for human occupied vehicle in three-dimensional underwater workspace | |
Deng et al. | Adaptive observer based backstepping controller design for dynamic ship positioning | |
Peng et al. | Research on hover control of AUV uncertain stochastic nonlinear system based on constructive backstepping control strategy | |
Miao et al. | DOPH∞-based path-following control for underactuated marine vehicles with multiple disturbances and constraints | |
VanZwieten et al. | Design of a prototype ocean current turbine—Part II: flight control system | |
Bańka et al. | Design of a multivariable neural controller for control of a nonlinear MIMO plant | |
Brodtkorb et al. | Increasing the operation window for dynamic positioned vessels using the concept of hybrid control | |
CN113671977B (en) | Synchronous stable robust control method for state of offshore operation ship | |
CN112327638B (en) | Unmanned ship trajectory tracking optimal control method with designated performance and input saturation limitation | |
Ngongi et al. | Design of generalised predictive controller for dynamic positioning system of surface ships |
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 |