Performance optimization-based sail type ship wind power plant autonomous cruise control method

Abstract

The invention discloses a sail-type ship wind power plant autonomous cruise control method based on performance optimization, and the method comprises the steps: obtaining a position displacement error and a heading displacement error of a drum sailboat through introducing a displacement function, and achieving the autonomous cruise control of the drum sailboat based on an error conversion technology; obtaining a position change error of the drum sailboat considering the preset performance of the drum sailboat and a heading change error of the drum sailboat; and a main engine rotating speed controller and a rudder angle controller of the drum sailboat are designed in combination with a dynamic surface technology, the main engine rotating speed and the rudder angle of the drum sailboat are obtained, and autonomous cruise control over the sail type ship wind power plant is achieved. The method can meet the high tracking precision requirement of a narrow channel or a multi-obstacle water area, greatly improves the ship tracking precision, and meets the high tracking precision requirement of a complex water area. Meanwhile, the autonomous cruise task of the ship under the severe sea condition can be effectively guaranteed, a powerful guarantee is provided for guaranteeing the transient performance and the steady-state performance of the ship, and a key technical support is provided for safe navigation of the drum sailboat in a complex water area.

Description

Technical Field

[0001] The present invention relates to the technical field of ship motion control, and in particular to an autonomous cruise control method for a sail-type ship wind farm based on performance optimization. Background Art

[0002] In recent years, with the rapid growth of carbon emissions from global ocean transportation and other maritime activities, rotary sailboats are receiving more and more attention due to their unique energy-saving characteristics and other potential advantages. In the field of marine engineering, realizing autonomous path tracking control of rotary sailboats can effectively reduce ship fuel consumption, save operating costs, and meet the long-term requirements of specific navigation missions. This technology has broad application prospects in maritime search and rescue, offshore resource detection, offshore wind farm cruising and other fields. However, the existing path tracking control algorithm for sailing ships still has the following two defects:

[0003] Under the action of the drum sail, the impact of external environmental disturbances on the control system is more significant, and the time-varying ocean environment can easily lead to instability of the control system. The special coupling effect between the drum sail and the hull also increases the uncertainty of the system model to a certain extent. In addition, compared with traditional underdriven ships, drum sails face stronger model uncertainty and external interference, which poses a huge challenge to the stability and robustness of the closed-loop control algorithm. Although neural networks, fuzzy logic systems, etc. have good nonlinear approximation capabilities, they still cannot meet the robustness requirements of sailing ships. It is urgent to study a model parameter / structure uncertainty optimization method for high coupling characteristics.

[0004] Existing control algorithms are mainly suitable for cruising missions in open waters, and it is difficult to meet the high tracking accuracy requirements in narrow channels or waters with multiple obstacles. In particular, for special maritime scenarios such as offshore wind farm inspections, sailing ships should have quantitative performance-constrained navigation capabilities to achieve safe and efficient cruising missions. In addition, most control algorithms assume that the ship actuator gain function is known, which is somewhat different from actual marine engineering and may reduce the effectiveness and control accuracy of related theoretical algorithms, affecting the actual navigation safety of ships. Summary of the invention

[0005] The invention discloses a sail-type ship wind farm autonomous cruise control method based on performance optimization to overcome the above technical problems.

[0006] In order to achieve the above object, the technical solution of the present invention is:

[0007] A sail-type ship wind farm autonomous cruise control method based on performance optimization comprises the following steps:

[0008] S1: Establish a nonlinear mathematical model of a 3-DOF rotary sailboat;

[0009] S2: establishing a kinematic equation of a logical virtual ship, obtaining the position coordinates of the logical virtual ship, and obtaining the azimuth angle of the logical virtual ship relative to the real ship according to the nonlinear mathematical model of the 3-DOF rotary sailboat;

[0010] S3: by introducing a displacement function, obtaining a position displacement error and a bow displacement error of the rotary sailboat, so as to obtain a position transformation error and a bow transformation error of the rotary sailboat considering the preset performance of the rotary sailboat based on an error transformation technology;

[0011] S4: obtaining a virtual control law according to the azimuth of the logical virtual ship relative to the real ship, the position change error of the rotary sailboat and the bow change error of the rotary sailboat: obtaining a dynamic surface signal based on the dynamic surface control technology, and then obtaining the difference between the dynamic surface signal and the virtual control law;

[0012] S5: According to the dynamic surface signal, the difference between the dynamic surface signal and the virtual control law, the main engine speed controller and the rudder angle controller of the rotary sailboat based on the adaptive gain variable are obtained to obtain the main engine speed and rudder angle of the rotary sailboat, so as to realize the autonomous cruise control of the sailing ship wind farm.

[0013] Beneficial effects: The present invention is an autonomous cruise control method for a sailing ship wind farm based on performance optimization. By introducing a displacement function, the position displacement error and bow displacement error of a rotary sailboat are obtained. Based on the error transformation technology, the position transformation error and bow transformation error of the rotary sailboat considering the preset performance of the rotary sailboat are obtained; then the virtual control law is obtained in combination with the azimuth of the logical virtual ship relative to the real ship: the dynamic surface signal is obtained based on the dynamic surface control technology, and then the difference between the dynamic surface signal and the virtual control law is obtained, and the main engine speed controller and rudder angle controller of the rotary sailboat are designed to obtain the main engine speed and rudder angle of the rotary sailboat, so as to realize the autonomous cruise control of the sailing ship wind farm. The present invention achieves the high tracking accuracy requirements of narrow channels or waters with multiple obstacles by obtaining the position transformation error and bow transformation error of the rotary sailboat considering the preset performance of the rotary sailboat, and can greatly improve the tracking accuracy of the ship to meet the high tracking accuracy requirements of complex waters. At the same time, the design of the main engine speed controller and rudder angle controller of the rotary sailboat based on adaptive gain variables can effectively ensure the ship's autonomous cruising mission in severe sea conditions, provide a strong guarantee for ensuring the transient performance and steady-state performance of the ship, and provide key technical support for the safe navigation of the rotary sailboat in complex waters. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0015] Figure 1 is a flow chart of the autonomous cruise control method of the present invention;

[0016] Figure 2 It is a framework diagram of the LVS guidance principle in an embodiment of the present invention;

[0017] Figure 3 is a signal flow chart of a sailing ship control system in an embodiment of the present invention;

[0018] Figure 4 is a conceptual signal flow chart in an embodiment of the present invention;

[0019] Figure 5 It is a schematic diagram of the thrust principle of the rotary sailboat in an embodiment of the present invention;

[0020] Figure 6 A schematic diagram of a three-dimensional trajectory in an offshore wind farm cruise mission in an embodiment of the present invention;

[0021] Figure 7a A schematic diagram of position path tracking error in an embodiment of the present invention;

[0022] Figure 7b is a schematic diagram of a heading path tracking error in an embodiment of the present invention;

[0023] Figure 8a This is a schematic diagram comparing the control command and actual input of the host speed in an embodiment of the present invention;

[0024] Figure 8b Schematic diagram of comparison between the control command and actual input of the rudder angle in an embodiment of the present invention;

[0025] Fig. 9 is a drum speed curve of a sailing vessel in an embodiment of the present invention;

[0026] Fig.10a is a thrust optimization rate curve of a rotary sail in an embodiment of the present invention;

[0027] Fig.10b is an energy optimization rate curve in an embodiment of the present invention;

[0028] Fig.11 is a path tracking trajectory curve in an embodiment of the present invention;

[0029] Fig.12a A comparison diagram of path tracking position errors under different control algorithms in an embodiment of the present invention;

[0030] Figure 12b is a comparison diagram of path tracking heading errors under different control algorithms in an embodiment of the present invention;

[0031] Fig.13a : is the main engine speed control command curve under different control algorithms in the embodiment of the present invention;

[0032] Fig.13b ] are the rudder angle control command curves under different control algorithms in the embodiments of the present invention. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] This embodiment introduces a sail-type ship wind farm autonomous cruise control method based on performance optimization, such as Figure 1-Figure 4 As shown, the following steps are included:

[0035] S1: Establish a nonlinear mathematical model of a 3-DOF rotary sailboat;

[0036] Preferably, the nonlinear mathematical model of the 3-DOF rotary sailboat is established as shown in formula (1):

[0037]

[0038] Where: x, y, ψ represent the horizontal coordinate, vertical coordinate and heading angle of the rotary sailboat in the earth-fixed coordinate system respectively; Ο represents the position vector of the rotary sailboat in the appendage coordinate system, where Ο=[u,v,r] T , u, v, r represent the forward speed, drift speed and bow speed of the rotary sailboat in the appendage coordinate system respectively; ψ represents the bow angle of the rotary sailboat; m u ,m v ,m r They represent the hydrodynamic additional mass of the rotary sailboat in the forward, horizontal and bow directions respectively; f u (Ο),f v (Ο),f r (Ο) represents the model uncertainty of the rotary sailboat in the forward, sideways and bow pitching degrees of freedom; τ up =Tu (·)|n|n represents the forward thrust generated by the propeller and rudder; where T u (·) represents the host gain function, n represents the host speed, |n| represents the absolute value of the host speed; τ us represents the compensating thrust generated by the drum sail; d wu ,d wv ,d wr They represent the unknown time-varying environmental disturbances of the rotary sailboat in the forward, sideways and bow pitching degrees of freedom; τ rr =F r (·)δ represents the bow moment generated by the propeller and rudder; d u1 ,d u2 ,d u3 ,d v1 ,d v2 ,d v3 ,d r1 ,d r2 ,d r3 All represent unknown hydrodynamic parameters, among which F r (·) represents the steering gear gain function, δ represents the rudder angle, which is used to describe the unknown hydrodynamic parameters; |u| represents the absolute value of the forward speed;

[0039] in,

[0040]

[0041] Where: A Indicates air density; V a represents the apparent wind speed; A represents the maximum projected area of ​​the rotary sail; C L ,C D They represent the lift and drag coefficients of the drum sail respectively; β represents the angle between the apparent wind direction and the bow direction of the ship.

[0042] S2: establishing a kinematic equation of a logical virtual ship, obtaining the position coordinates of the logical virtual ship, and obtaining the azimuth angle of the logical virtual ship relative to the real ship according to the nonlinear mathematical model of the 3-DOF rotary sailboat;

[0043] Specifically, in navigation practice, the reference path is usually generated by a series of waypoints set manually, such as W1, W2, W3..., W n Therefore, this embodiment proposes a waypoint-based logical virtual ship (LVS) guidance principle to provide a smooth reference path for the rotary sailboat and avoid the overshoot phenomenon of traditional line-of-sight guidance.

[0044] Preferably, the azimuth of the logical virtual ship relative to the real ship is obtained as follows:

[0045] S21: The kinematic equation of LVS is established as formula (3):

[0046]

[0047] In the formula, x d ,y d Respectively represent the horizontal coordinate and vertical coordinate of the LVS position; u d represents the expected speed of LVS; r d =u d / R i represents the desired bow angular velocity of LVS, R i represents the turning radius at the i-th waypoint; ψ d Indicates the heading angle of LVS;

[0048] S22: Get the azimuth of LVS relative to the actual ship as follows:

[0049]

[0050] Where: r represents the azimuth of LVS relative to the real ship; sgn(·) represents the sign function.

[0051] Next, a rotor sailboat path tracking controller is established to complete the autonomous cruise control of the sailing ship wind farm:

[0052] S3: By introducing a displacement function, the position displacement error and the bow displacement error of the rotary sailboat are obtained, so as to obtain the position transformation error and the bow transformation error of the rotary sailboat based on the error transformation technology;

[0053] Specifically, the logical virtual ship guidance principle based on waypoints can effectively reduce human intervention, improve the execution efficiency of marine cruise tasks, and play a vital role in the autonomous navigation of sailing ships. In terms of controller design, this embodiment comprehensively uses preset performance control, reinforcement learning technology, adaptive technology and dynamic surface control (DSC) technology to design a rotary sailboat path tracking controller. The design process is mainly divided into the following two steps:

[0054] Preferably, the method for obtaining the position change error and the heading change error of the rotary sailboat is as follows:

[0055] S31: First, introduce the following shift function ξ i :

[0056]

[0057] Where: i represents the shift function; T srepresents the convergence time set for the displacement function; t represents the time; n represents the adjustment parameter of the slope of the displacement function; where z represents the displacement variable of the rotary sailboat;

[0058] S32: Performing a primary transformation on the output error according to the displacement function to obtain the position displacement error and bow displacement error of the rotary sailboat, as shown in formula (6):

[0059]

[0060] In the formula, z e represents the position error of the drum sailboat from the LVS; ψ e represents the heading error between the heading angle of the rotary sailboat and the expected heading angle; ξ z The shift function representing the position error, ξ ψ represents the shift function of the heading error; Indicates position shift error; Indicates heading shift error.

[0061] S33: According to the position displacement error and the heading displacement error of the rotary sailboat, the output error is transformed twice by the error transformation technology to obtain the position transformation error and the heading transformation error ε of the rotary sailboat z ,ε ψ .

[0062]

[0063] in,

[0064]

[0065] Where: ε z Represents the position change error of the rotary sailboat; ε ψ represents the heading change error of the rotary sailboat; ρ i represents the preset performance function; ρ i0 Represents the preset initial value; ρ if Indicates the maximum error boundary in the stable stage; l i is the convergence rate parameter; T i Indicates the preset convergence time for the performance function;

[0066] Therefore, in this embodiment, the error preset range R0 is defined as {ε i ||-ρ i <ε i < i}.

[0067] S4: In order to stabilize the position error and the heading error, a virtual control law is obtained according to the azimuth of the logical virtual ship relative to the real ship, the position change error of the rotary sailboat, and the heading change error of the rotary sailboat: a dynamic surface signal is obtained based on the dynamic surface control technology, and then a difference between the dynamic surface signal and the virtual control law is obtained;

[0068] Preferably, the method for obtaining the difference between the dynamic surface signal and the virtual control law is as follows:

[0069] S41: The virtual control law is obtained as shown in formula (9).

[0070]

[0071] Where: α u represents the virtual control law on the forward degree of freedom; α r represents the virtual control law on the bow pitch degree of freedom; represents the virtual design parameters for forward and yaw degrees of freedom; ξ z represents the displacement function on the forward degree of freedom; ξ ψ represents the displacement function of the yaw degree of freedom; ρ z represents the preset performance function on forward freedom; ρ ψ represents the preset performance function on the free pitch of the bow; ε z Represents the position change error of the rotary sailboat; ε ψ Indicates the heading change error of the rotary sailboat; x d ,y d Respectively represent the horizontal coordinate of the position of LVS; ψ r Indicates the azimuth of LVS relative to the actual ship; z e represents the position error of the drum sailboat from the LVS; ψ e represents the heading error between the heading angle of the rotary sailboat and the desired heading angle; v represents the transverse drift speed of the rotary sailboat in the appendage coordinate system; Indicates position shift error; Indicates the heading shift error; represents the first-order derivative.

[0072] S42: According to the virtual control law, based on the dynamic surface control technology, a dynamic surface signal is obtained to obtain a difference between the dynamic surface signal and the virtual control law;

[0073] Specifically, in the traditional backstepping design process, due to the need to u and α rWhen performing repeated differentiation, the problem of "computational explosion" is inevitable. In order to solve this problem, the dynamic surface control technology (DSC) is introduced to reduce the calculation dimension and simplify the subsequent controller design.

[0074] Preferably, the formula used to obtain the dynamic surface signal is as follows:

[0075]

[0076] In the formula, θ u ,θ r They represent the time constants greater than zero for the forward and yaw degrees of freedom, respectively; β u represents the dynamic surface signal on the forward degree of freedom; β r Represents the dynamic surface signal on the yaw degree of freedom; β u (0) represents the initial dynamic surface signal on the forward degree of freedom; α u (0) represents the initial virtual control law on the forward degree of freedom; β r (0) represents the initial dynamic surface signal on the yaw degree of freedom; α r (0) represents the initial virtual control law on the yaw degree of freedom;

[0077] Therefore, the difference between the dynamic surface signal and the virtual control law is obtained as follows:

[0078] y u =α u -β u ,y r =α r -β r

[0079] Where: y u Represents the dynamic surface error in the forward degree of freedom; y r Indicates the dynamic surface error on the yaw degree of freedom;

[0080] S5: According to the dynamic surface signal, the difference between the dynamic surface signal and the virtual control law, the main engine speed controller and the rudder angle controller of the rotary sailboat are obtained to obtain the main engine speed and rudder angle of the rotary sailboat, so as to realize autonomous cruise control of the sailing ship wind farm.

[0081] Preferably, the method for obtaining the main engine speed controller and the rudder angle controller of the rotary sailboat is as follows:

[0082] S51: Define the dynamic error u e =β u -u,r e =β r -r, then the derivative is as follows:

[0083]

[0084] Where: u e The dynamic error of the forward speed of the drum sailboat; r e represents the dynamic error of the bow angular velocity of the drum sail; τ us represents the compensating thrust generated by the rotor sail; τ up represents the forward thrust produced by the propeller and rudder; Figure 5 The schematic diagram of the thrust principle of the rotary sailboat is shown; τ rr Indicates the bow moment generated by the propeller and rudder; m u ,m r They represent the hydrodynamic additional mass of the rotary sailboat in the forward and bowing directions respectively; f uu (Ο u ),f rr (Ο r ) represent the combined uncertainties in the forward and forward degrees of freedom respectively:

[0085] f uu (Ο u )=-m v vr+f u (Ο)-d wu

[0086] f rr (Ο r )=-(m u -m v )uv+f r (Ο)-d wr (12)

[0087] In the formula

[0088] The combined uncertainty term f on the forward and yaw degrees of freedom is calculated by the actor neural network. uu (Ο u ),f rr (Ο r ) is used for online approximation, as shown in formula (13): v are the hydrodynamic additional mass of the drum-shaped sailboat in the transverse drift direction; u, v, r are the forward speed, transverse drift speed and bow rolling speed of the drum-shaped sailboat in the appendage coordinate system; d wu ,d wr They represent the unknown time-varying environmental disturbances on the forward and bowing degrees of freedom of the rotary sailboat respectively;

[0089] Specifically, uu (Ο u ),f rr (Ο r ) represents the combined uncertainty term,

[0090]

[0091]

[0092] Where W au ,W ar Represents the weights of the actor's neural network in terms of displacement speed and direction; S au (Ο u ),S ar (Ο r ) represent the Gaussian functions of the forward freedom and the bow freedom respectively; ∈(Ο u ),∈(Ο r ) represent the approximation errors of the forward degree of freedom and the yaw degree of freedom respectively; u ,Ο r They represent the set of input vectors of forward degree of freedom and yaw degree of freedom respectively; T represents the transposition sign.

[0093] S52: According to the dynamic error, the dynamic surface signal, and the difference between the dynamic surface signal and the virtual control law, the main engine speed controller and the rudder angle controller of the rotary sailboat are obtained as follows:

[0094] Specifically, this embodiment combines the adaptive compensation technology to design controllers for the main engine speed n and the rudder angle δ, as shown in equations (14) and (15):

[0095]

[0096] Where: N = |n|n represents the intermediate variable related to the main engine speed; n represents the main engine speed, δ represents the rudder angle; λ u represents the gain design variable on the forward degree of freedom; λ r represents the gain design variable on the yaw degree of freedom; Represents λ r An estimated value of Represents λ u The estimated value of α N Indicates the control input related to the main engine speed; a δ Indicates the control input related to the main engine rudder angle; ι u1 ,ι u2 ,ι r1 ,ι r2 Both represent positive controller design parameters; u e The dynamic error of the forward speed of the drum sailboat; r e W represents the dynamic error of the bow angular velocity of the rotary sailboat; au represents the weight of the actor neural network on the forward degree of freedom; W arrepresents the weight of the actor's neural network on the pitching degree of freedom; W au An estimated value of W ar The estimated value of τ us represents the compensating thrust generated by the rotor sail; express The initial value of express The initial value of T u (·) represents the host gain function, F r (·) represents the servo gain function; express The first-order derivative of is the gain adaptation rate on the displacement velocity; express The first-order derivative of is the gain adaptation rate in the displacement direction;

[0097] in,

[0098]

[0099] Where: express The first derivative of ; express The first derivative of au represents the learning rate of the actor neural network on the forward degrees of freedom, σ ar K represents the learning rate of the actor’s neural network on the pitching degree of freedom; cu ,k cr W represents the weight parameter designed for forward and yaw degrees of freedom; cu represents the critic neural network weight on displacement velocity, W cr represents the critic neural network weight in the displacement direction; S cu represents the Gaussian function on displacement velocity, S cr represents the Gaussian function in the displacement direction; W cu An estimated value of W cr An estimated value of

[0100] in,

[0101]

[0102] Where: cu represents the learning rate of the critic neural network on the forward degrees of freedom, σ cr represents the learning rate of the critic neural network on the bow DOF; p u ,p r, respectively represent the designed transformation parameters for forward and yaw degrees of freedom; q u ,q r They represent the design input parameters for forward and yaw degrees of freedom respectively; ε z Represents the position change error of the rotary sailboat; ε ψ Indicates the heading change error of the rotary sailboat; τ up represents the forward thrust generated by the propeller and rudder; τ rr represents the bow moment generated by the propeller and rudder Represents ε z The first derivative of ; Represents ε ψ The first derivative of ; Indicates S cu Find partial derivatives; Indicates S cr Find partial derivatives; denote the future discount factors of the critic neural network on the forward and yaw degrees of freedom, respectively;

[0103] Specifically, in order to verify the effectiveness and superiority of this embodiment, based on the established drum sailing model, a 3×4 row and column layout offshore wind farm was selected to conduct a path tracking experiment in a simulated ocean environment. First, according to the cruising mission requirements of the wind farm, reasonable waypoints were selected near the wind turbines. A smooth reference path was generated through the logical virtual ship, and the LVS expected speed u d The speed is set to 6m / s. The scanning radius of the rotary sailboat is set to 60m. It is worth noting that the selection of waypoints, the expected speed and the scanning radius are all determined by the user, which improves the practicality and effectiveness of the proposed algorithm and is more conducive to its application in marine engineering practice.

[0104] Figure 6 The three-dimensional trajectory of the offshore wind farm cruise mission is shown. Through the preset waypoint information, the scanning range of the rotary sailboat can effectively cover all wind turbines in the wind farm and realize autonomous detection of the seabed geology near the wind turbine pile foundation. Figure 7a and Figure 7b It represents the path tracking error under the proposed algorithm, which reflects the excellent transient and steady-state performance of the controller. As can be seen from the figure, although the initial value of the output error does not meet the tracking condition, the proposed algorithm can still ensure that the tracking error converges to the preset area, which reflects the advantage of integrating the shift function into the PPC controller. Figure 8a and Figure 8b It represents the control command and actual input under the proposed algorithm. Due to the use of the steering gear servo system, the actual input of the host and the rudder angle n p ,δ r All are within a reasonable range and consistent with engineering practice. Fig. 9It represents the rotor speed of the rotor sailboat under the algorithm of this embodiment, indicating that its speed can be dynamically adjusted within a reasonable range to optimize the sailing thrust. Fig.10a and Fig.10b The thrust and energy optimization rate (EOR) curves of the rotary sail are shown respectively, EOR(t) = τ us (t) / (τ up (t)+τ us (t)). The calculation results show that the use of the drum sailing strategy can save about 12% of energy consumption.

[0105] Figure 11-13b The experimental results comparing the algorithm of this embodiment with the prior art are shown. Fig.11 and Fig.12a and Figure 12b It can be seen that although both algorithms can eventually converge to the reference path, the algorithm in this embodiment demonstrates higher control accuracy and smoother ship steering by virtue of its superiority in reinforcement learning technology and preset performance control. Fig.13a and Fig.13b The control command curves obtained under the two control algorithms are shown. Due to the constraint function of the performance function, the algorithm of this embodiment makes the control command curve smoother, effectively reduces the vibration of the actuator, and is of great significance to the stable navigation of the rotary sailboat.

[0106] Combining the prior art, guidance law construction, controller design and simulation experiment comparison, the present invention has the following two significant beneficial effects in sailing ship path tracking:

[0107] This embodiment is based on the autonomously modeled drum sailboat model, and the proposed reinforcement learning optimization algorithm based on actor-critic neural networks (AC-NNs), in which the actor neural network is responsible for interacting with the environment and generating appropriate control strategies, while the critic neural network is responsible for evaluating the performance of the actor to guide its actions in the next stage, thereby minimizing the output error. The algorithm effectively improves the reliability and robustness of the control system, can effectively solve the problems of uncertainty in the ship system model and interference from the external marine environment during the autonomous cruising of sailing ships in wind farms, and significantly reduces the negative impact of the drum sail equipment on the stability of the ship. Therefore, this embodiment can effectively realize the long-duration autonomous cruising mission of the drum sailboat in harsh marine environments, and improve the reliability and economy of the sailing of sailing ships.

[0108] Taking into account marine engineering practice, this embodiment designs a preset performance control algorithm that is not constrained by the initial state of the ship by introducing a shift function. The algorithm simultaneously guarantees the transient performance and steady-state performance of the sailing ship, significantly improves the control accuracy and response speed of the sailing ship, and meets the navigation mission requirements in complex or restricted waters, providing key technical support for the safe navigation of the rotary sailing ship in complex waters. In addition, adaptive technology is used to compensate the actuator gain function online, reducing the gain change caused by actuator wear, and further improving the control accuracy, which is of great value to ensuring the navigation safety of sailing ships and promoting the application of rotary sailing ships in engineering practice.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sail-type ship wind farm autonomous cruise control method based on performance optimization, characterized in that: The steps include: S1: Establish a nonlinear mathematical model of a 3-DOF rotary sailboat; S2: establishing a kinematic equation of a logical virtual ship, obtaining the position coordinates of the logical virtual ship, and obtaining the azimuth angle of the logical virtual ship relative to the real ship according to the nonlinear mathematical model of the 3-DOF rotary sailboat; S3: by introducing a displacement function, obtaining a position displacement error and a bow displacement error of the rotary sailboat, so as to obtain a position transformation error and a bow transformation error of the rotary sailboat considering the preset performance of the rotary sailboat based on an error transformation technology; S4: obtaining a virtual control law according to the azimuth of the logical virtual ship relative to the real ship, the position change error of the rotary sailboat and the bow change error of the rotary sailboat: obtaining a dynamic surface signal based on the dynamic surface control technology, and then obtaining the difference between the dynamic surface signal and the virtual control law; S5: According to the dynamic surface signal, the difference between the dynamic surface signal and the virtual control law, the main engine speed controller and the rudder angle controller of the rotary sailboat based on the adaptive gain variable are obtained to obtain the main engine speed and rudder angle of the rotary sailboat, so as to realize the autonomous cruise control of the sailing ship wind farm.

2. The method for autonomous cruise control of a sailing ship wind farm based on performance optimization according to claim 1 is characterized in that: In S5, the method for obtaining the main engine speed controller and the rudder angle controller of the rotary sailboat is as follows: S51: Define the dynamic error u e =β u -u,r e =β r -r, then the derivative is as follows: Where: u e The dynamic error of the forward speed of the drum sailboat; r e represents the dynamic error of the bow angular velocity of the drum sail; τ us represents the compensating thrust generated by the rotor sail; τ up represents the forward thrust generated by the propeller and rudder; τ rr Indicates the bow moment generated by the propeller and rudder; m u ,m r They represent the hydrodynamic additional mass of the rotary sailboat in the forward and bowing directions respectively; f uu (Ο u ),f rr (Ο r ) represent the combined uncertainty terms of the forward and forward degrees of freedom respectively: β u Represents the dynamic surface signal on the forward degree of freedom; β r Indicates the dynamic surface signal on the yaw degree of freedom; f uu (O u )=-m v vr+f u (O)-d wu f rr (O r )=-(m u -m v )uv+f r (O)-d wr Where: m v are the hydrodynamic additional mass of the drum-shaped sailboat in the transverse drift direction; u, v, r are the forward speed, transverse drift speed and bow rolling speed of the drum-shaped sailboat in the appendage coordinate system; d wu ,d wr They represent the unknown time-varying environmental disturbances on the forward and bowing degrees of freedom of the rotary sailboat; f u (Ο),f r (Ο) represents the model uncertainty of the rotary sailboat in the forward and bow pitch degrees of freedom; The combined uncertainty term f on the forward and yaw degrees of freedom is calculated by the actor neural network. uu (Ο u ),f rr (Ο r ) performs online approximation as follows: Where W au ,W ar Respectively represent the weights of the actor's neural network in terms of displacement speed and direction; S au (Ο u ),S ar (Ο r ) represent the Gaussian functions of the forward freedom and the bow freedom respectively; ∈(Ο u ),∈(Ο r ) represent the approximation errors of the forward degree of freedom and the yaw degree of freedom respectively; u ,Ο r represents the set of input vectors of forward freedom and yaw freedom respectively; T represents the transposition symbol; y u Represents the dynamic surface error in the forward degree of freedom; y r Indicates the dynamic surface error on the yaw degree of freedom; S52: According to the dynamic error, the dynamic surface signal, and the difference between the dynamic surface signal and the virtual control law, the formulas used for the main engine speed controller and the rudder angle controller of the rotary sailboat are obtained as follows: Where: N = |n| n represents the intermediate variable related to the main engine speed; n represents the main engine speed, δ represents the rudder angle; λ u represents the gain design variable on the forward degree of freedom; λ r represents the gain design variable on the yaw degree of freedom; Represents λ r An estimated value of Represents λ u The estimated value of N Represents the control input related to the main engine speed; α δ Indicates the control input related to the main engine rudder angle; ι u1 ,ι u2 ,ι r1 ,ι r2 Both represent positive controller design parameters; u e The dynamic error of the forward speed of the drum sailboat; r e W represents the dynamic error of the bow angular velocity of the rotary sailboat; au represents the weight of the actor neural network on the forward degree of freedom; W ar represents the weight of the actor's neural network on the pitching degree of freedom; W au An estimated value of W ar The estimated value of τ us represents the compensating thrust generated by the rotor sail; express The initial value of express The initial value of T u (·) represents the host gain function, F r (·) represents the servo gain function; express The first-order derivative of is the gain adaptation rate on the displacement velocity; express The first-order derivative of is the gain adaptation rate in the displacement direction; in, Where: express The first derivative of ; express The first derivative of au represents the learning rate of the actor neural network on the forward degrees of freedom, σ ar K represents the learning rate of the actor’s neural network on the pitching degree of freedom; cu ,K cr W represents the weight parameter designed for forward and yaw degrees of freedom; cu represents the critic neural network weight on displacement velocity, W cr represents the critic neural network weight in the displacement direction; S cu represents the Gaussian function on displacement velocity, S cr represents the Gaussian function in the displacement direction; W cu An estimated value of W cr An estimated value of in, Where: cu represents the learning rate of the critic neural network on the forward degrees of freedom, σ cr represents the learning rate of the critic neural network on the bow DOF; p u ,p r , respectively represent the designed transformation parameters for forward and yaw degrees of freedom; q u ,q r They represent the design input parameters for forward and yaw degrees of freedom respectively; ε z Represents the position change error of the rotary sailboat; ε ψ Indicates the heading change error of the rotary sailboat; τ up represents the forward thrust generated by the propeller and rudder; τ rr represents the bow moment produced by the propeller and rudder; Represents ε z The first derivative of ; Represents ε ψ The first derivative of ; Indicates S cu Find partial derivatives; Indicates S cr Find partial derivatives; denote the future discount factors of the critic neural network on the forward and yaw degrees of freedom, respectively.

3. The method for autonomous cruise control of a sailing ship wind farm based on performance optimization according to claim 1 is characterized in that: The nonlinear mathematical model of the 3-DOF rotary sailboat is established as follows: Where: x, y, ψ represent the horizontal coordinate, vertical coordinate and heading angle of the rotary sailboat in the earth-fixed coordinate system respectively; Ο represents the position vector of the rotary sailboat in the appendage coordinate system, where Ο=[u,v,r] T , u, v, r represent the forward speed, transverse drift speed and bow rolling speed of the rotary sailboat in the appendage coordinate system respectively; m u ,m v ,m r They represent the hydrodynamic additional mass of the rotary sailboat in the forward, horizontal and bow directions respectively; f u (Ο),f v (Ο),f r (Ο) represents the model uncertainty of the rotary sailboat in the forward, sideways and bow pitching degrees of freedom; τ up =T u (·)|n|n represents the forward thrust generated by the propeller and rudder; where T u (·) represents the host gain function, n represents the host speed, |n| represents the absolute value of the host speed; τ us represents the compensating thrust generated by the drum sail; d wu ,d wv ,d wr They represent the unknown time-varying environmental disturbances of the rotary sailboat in the forward, sideways and bow pitching degrees of freedom; τ rr =F r (·)δ represents the bow moment generated by the propeller and rudder; d u1 ,d u2 ,d u3 ,d v1 ,d v2 ,d v3 ,d r1 ,d r2 ,d r3 All represent unknown hydrodynamic parameters, among which F r (·) represents the steering gear gain function, δ represents the steering angle; |u| represents the absolute value of the forward speed; in, Where: A Indicates air density; V a represents the apparent wind speed; A represents the maximum projected area of ​​the rotary sail; C L ,C D They represent the lift and drag coefficients of the drum sail respectively; β represents the angle between the apparent wind direction and the bow direction of the ship.

4. The method for autonomous cruise control of a sailing ship wind farm based on performance optimization according to claim 1 is characterized in that: In S2, the azimuth of the logical virtual ship relative to the real ship is obtained as follows: S21: The kinematic equation of LVS is established as follows: In the formula, x d ,y d Respectively represent the horizontal coordinate and vertical coordinate of the LVS position; u d represents the expected speed of LVS; r d =u d / R i represents the desired bow angular velocity of LVS, R i represents the turning radius at the i-th waypoint; ψ d Indicates the heading angle of LVS; S22: Get the azimuth of LVS relative to the actual ship as follows: Where: r represents the azimuth of LVS relative to the real ship; sgn(·) represents the sign function.

5. The method for autonomous cruise control of a sailing ship wind farm based on performance optimization according to claim 1 is characterized in that: In S3, the method for obtaining the position change error and the heading change error of the rotary sailboat is as follows: S31: First, introduce the following shift function ξ i : Where: i represents the shift function; T s represents the convergence time set for the displacement function; t represents the time; n represents the adjustment parameter of the slope of the displacement function; where z represents the displacement variable of the rotary sailboat; S32: According to the displacement function, the position displacement error and the bow displacement error of the rotary sailboat are obtained: In the formula, z e represents the position error of the drum sailboat from the LVS; ψ e represents the heading error between the heading angle of the rotary sailboat and the expected heading angle; ξ z The shift function representing the position error, ξ ψ represents the shift function of the heading error; Indicates position shift error; Indicates the heading shift error; S33: Obtaining the position change error of the rotary sailboat and the bow change error ε of the rotary sailboat according to the position shift error and the bow shift error of the rotary sailboat z ,ε ψ as follows: in, Where: ε z Represents the position change error of the rotary sailboat; ε ψ represents the heading change error of the rotary sailboat; ρ i represents the preset performance function; ρ i0 Represents the preset initial value; ρ if Indicates the maximum error boundary in the stable stage; l i is the convergence rate parameter; T i represents the preset convergence time for the performance function; i represents an intermediate parameter.

6. The method for autonomous cruise control of a sailing ship wind farm based on performance optimization according to claim 1 is characterized in that: In S4, the difference between the dynamic surface signal and the virtual control law is obtained as follows: S41: Obtain the virtual control law as follows: Where: α u represents the virtual control law on the forward degree of freedom; α r represents the virtual control law on the bow pitch degree of freedom; represents the virtual design parameters for forward and yaw degrees of freedom; ξ z represents the displacement function on the forward degree of freedom; ξ ψ represents the displacement function of the yaw degree of freedom; ρ z represents the preset performance function on forward freedom; ρ ψ represents the preset performance function on the free pitch of the bow; ε z Represents the position change error of the rotary sailboat; ε ψ Indicates the heading change error of the rotary sailboat; x d ,y d Respectively represent the horizontal coordinate of the position of LVS; ψ r Indicates the azimuth of LVS relative to the actual ship; z e represents the position error of the drum sailboat from the LVS; ψ e represents the heading error between the heading angle of the rotary sailboat and the desired heading angle; v represents the transverse drift speed of the rotary sailboat in the appendage coordinate system; Indicates position shift error; Indicates the heading shift error; represents the first-order derivative; S42: According to the virtual control law, based on the dynamic surface control technology, a dynamic surface signal is obtained to obtain a difference between the dynamic surface signal and the virtual control law.

7. The method for autonomous cruise control of a sailing ship wind farm based on performance optimization according to claim 6 is characterized in that: In S42, the formula used to obtain the dynamic surface signal is as follows: In the formula, θ u ,θ r They represent the time constants greater than zero for the forward and yaw degrees of freedom, respectively; β u Represents the dynamic surface signal on the forward degree of freedom; β r Represents the dynamic surface signal on the yaw degree of freedom; β u (0) represents the initial dynamic surface signal on the forward degree of freedom; α u (0) represents the initial virtual control law on the forward degree of freedom; β r (0) represents the initial dynamic surface signal on the yaw degree of freedom; α r (0) represents the initial virtual control law on the yaw degree of freedom; Therefore, the difference between the dynamic surface signal and the virtual control law is obtained as follows: y u =a u -b u ,y r =a r -b r Where: y u Represents the dynamic surface error in the forward degree of freedom; y r Represents the dynamic surface error on the yaw degree of freedom.

Images