JPH09226689A - Automatic steering device for ship - Google Patents

Abstract

PROBLEM TO BE SOLVED: To ensure stable control of an automatic steering system by combining a general control object with uncertainty with an auto-pilot reach a H ∞ control problem so that a H ∞ norm for transmitting property from a disturbance to an estimation amount is not more than a specification. SOLUTION: When a set course ≃c is input, a deviation between the set cource ≃c and a stem azimuth ϕ is found by the first adder 11 and an instructed steering angle Uc is output from an auto-pilot 12 in accordance with the deviation and supplied to the second adder 13. The adder 13 subtracts a disturbance (d) converted into an angle from the instructed steering angle Uc and outputs the result to a control object 14. The third adder 15 subtracts a disturbance dv detected by a stem azimuth detector from an output signal ϕ0 from the control object 14 to produce a stem azimuth ϕ. That is, a general control object containing a hull and an azimuth detector is combined with the auto-pilot to reach a H ∞control problem so that a H ∞ norm for transmitting property from a disturbance to an estimation amount is not more than a specification.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control system for an automatic steering system for ships, and more particularly to a control system based on the H∞ control theory.

[0002]

2. Description of the Related Art FIG. 6 shows a block diagram of a control system of a conventional marine vessel automatic steering system. Hereinafter, such a control system will be simply referred to as an automatic steering system. The automatic steering system includes an automatic steering device, that is, an autopilot 120, a steering device 16, a hull 14-1, and a heading detector 14-2.

The automatic steering device inputs a set course φ _C and a heading φ, and outputs a command rudder angle U _C. This command steering angle U
_C is supplied to the steering gear 16. The steering gear 16 has a command steering angle U
Input _C and output steering angle U. The steering gear 16 has a servo mechanism for causing the rudder angle U to quickly follow the command rudder angle U _C , and corresponds to a first-order lag element.

A disturbance d such as wind and waves acts on the hull 14-1. The adder 13 adds the disturbance d converted into an angle to the steering angle U which is the output signal of the steering device 16. The hull 14-1 can be considered as a motion system that rotates around the azimuth axis of the ship. The hull 14-1 inputs the output signal U + d of the adder 13 and outputs the angular velocity dφ / dt in the bow direction.

The heading detector 14-2 calculates the heading φ from the angular velocity dφ / dt of the heading. The heading detector 14-2 may include a gyro compass, a magnetic compass, or the like. Such a heading φ is fed back to the autopilot 120. In this way, a stable closed loop is formed, and the heading φ follows the set course φ _C.

Automatic steering device or autopilot 120
The configuration and operation of will be described. The autopilot 120 has two adders 120-1, 120-4 and a proportional gain K _P.
And two coefficient multipliers 12 each having a differential gain T _D
It has 0-2, 120-3 and a Kalman filter 120-5.

The first adder 120-1 calculates the deviation ERR = φ _C −φ _E between the estimated value φ _E of the bow direction output from the Kalman filter 120-5 and the set direction φ _C. The function of autopilot 120 is to reduce this deviation ERR to zero.

The deviation ERR output from the first adder 120-1 is supplied to the second adder 120-4 via the coefficient unit 120-2 of the proportional gain K _P. The estimated value (dφ / dt) _E of the turning angular velocity dφ / dt output from the Kalman filter 120-5 is the coefficient unit 12 of the differential gain T _D.
It is supplied to the second adder 120-4 via 0-3. The second adder 120-4 adds the two output signals to calculate the command steering angle U _C. Commanded steering angle U _C is steering gear 16
And the Kalman filter 120-5.

The Kalman filter 120-5 has a heading φ
And the command steering angle U _C are input to calculate an estimated value φ _{E of the} heading φ and an estimated value (dφ / dt) _E of the turning angular velocity.

The Kalman filter 120-5 has a model of an object to be controlled (here, it is composed of a hull 14-1 and a heading detector 14-2), and the estimated value φ _{E of the} heading and a turning angular velocity are included as state quantities. It has an estimated value (dφ / dt) _E. The Kalman filter 120-5 has a feedback mechanism such that the estimated heading φ _E matches the heading φ, and calculates the estimated heading φ _E as a value from which noise components included in the heading φ have been removed. Output. Furthermore, the turning angular velocity dφ / d that cannot be directly obtained
As t, an estimated value (dφ / d) of the turning angular velocity which is the state quantity
t) _E is calculated and output. Kalman filter 120-
The reason why the command steering angle U _C is input to 5 is to match the model of the controlled object.

The noise component removal characteristic is determined by the Kalman filter constant. Kalman filter 120-
Since the detailed description of No. 5 and the method of obtaining the Kalman filter constant are well known in the field of control engineering, detailed description will be omitted.

Next, the control gain of the automatic steering device 120 is
Proportional gain K_PAnd differential gain T _DExplain about
You. Generally, the control gain is the following evaluation function J_APThe minimum
It is calculated based on the optimal control law.

[0013]

[Equation 1] J _AP = ∫ ₀ (φ ² + λU _c ² ) dt

Here, ∫ ₀ is the integral from 0 to ∞ with respect to time t, φ is the heading of the boat, U _C is the command steering angle, and λ is the weight.

The evaluation function J _AP is calculated from the heading φ and the command rudder angle U
_Since the balance of _C can be set, it can also be used for economic evaluation of automatic steering systems. The proportional gain K _P and the differential gain T can be obtained by solving the equation (1).
You can ask for _D. This solution is shown in the following document (1), for example.

Automatic steering device or autopilot 120
Are generally operated in a needle change mode and a needle hold mode. The needle change mode is different from the previous setting, and the set course φ _C
Corresponds to the time when the needle is changed, and the needle keeping mode corresponds to the time when the needle is kept to keep the previous setting state.

In the variable needle mode, the autopilot 120 does not include an integral operation in order to improve the responsiveness. on the other hand,
In the needle keeping mode, an integration operation is included in order to remove an error in the bow direction φ caused by the bias component included in the disturbance d. The autopilot 120 described with reference to FIG. 6 above is for the needle changing mode.

The construction and operation of the automatic steering device for the needle keeping mode, that is, the autopilot 120 'will be described with reference to FIG. Compared with the autopilot 120 for the needle changing mode shown in FIG.
The configuration is similar except that an integrator 120-6 having an integration time constant T _I is added.

The deviation ERR output from the first adder 120-1 is passed through the integrator 120-6 to the second adder 1
20-4. The output signal of the integrator 120-6 corresponds to the bias component included in the disturbance d. Therefore the second
The command steering angle U _C output from the adder 120-4 of
Since the component that cancels the bias component included in the disturbance d is included, the bow direction φ consistently matches the set direction φ _C.

Although the autopilots 120 and 120 'described with reference to FIGS. 6 and 7 are designed in consideration of the target ship, the characteristics of the actual ship may differ from those at the time of designing.
The characteristics may change due to the influence of cargo, etc., and the actual ship may not be able to exhibit the designed performance.

FIG. 8 shows a needle change response characteristic of a conventional autopilot. The vertical axis represents the (azimuth angle) needle change amount, and the horizontal axis represents time. The set course φ _C at the time of keeping the needle is set to zero, and the time point t
= 0, auto-pilot stepwise set course φ
_It is assumed that _C is input and the command steering angle U _C is output.
This command rudder angle U _C is supplied to the control unit 16 and the heading φ of the hull 14-1 changes accordingly.

The straight line 125 represents the set course φ _C , the solid curve 126 represents the designed response characteristics, and the broken line 127,
A dotted line 128 and a chain line 129 represent actual response characteristics. As is clear from comparison between the curve 126 and the curves 127, 128, and 129, it can be seen that the actual response characteristic is considerably different from the designed response characteristic.

In such a case, the operator adjusts the above-mentioned control gains T _I , K _P and T _D and the Kalman filter constant so as to match the actual characteristics of the ship. This adjustment work is possible even when the needle is held, but it is mainly performed when the needle is changed because the amount of change in the bow direction φ is smaller when the needle is held than when the needle is changed. This adjustment work is time consuming, troublesome, and requires skill.

[0024]

The conventional autopilots 120 and 120 'have the following problems. (1) When considering the characteristics of the ship to be controlled, changes in the characteristics are not explicitly incorporated in the design specifications.

(2) When setting the control gain and the Kalman filter constant, the characteristic change of the controlled object, that is, the parameter change cannot be explicitly taken in.

(3) Since the control gain and the Kalman filter constant are set independently, the performance of the autopilot cannot be clearly understood.

In view of these points, it is an object of the present invention to provide an autopilot that considers changes in the characteristics of a ship and has a clear performance against disturbance.

In view of these points, the present invention has an object to solve the problem of the automatic steering device by reducing it to the H ∞ control problem.

[0029]

According to the present invention, an automatic marine vessel configured to be closed-loop controlled by an automatic steering device which inputs a signal indicating a deviation of a bow direction with respect to a set course and outputs a commanded steering angle. In the steering device, the commanded steering angle is input, the heading is output, and the control target consisting of the hull and the azimuth detector is treated as a generalized control target with uncertainty, and the generalized control target and the automatic steering device are handled. Combined to reduce to the H ∞ control problem, the H ∞ control norm of the transfer characteristic from the disturbance of the generalized control target to the evaluation amount is set to be less than the specification, and the generalized control target having the above uncertainty is scaled using a constant ε. It is characterized in that it is converted into a generalized controlled object and reduced to an H∞ control problem.

According to the present invention, in the automatic steering system for a ship, the constant ε is an optimum value for minimizing the transfer characteristic of the H∞ control norm from the disturbance to the evaluation amount in the scaled generalized controlled object. It is characterized by having a set value.

According to the present invention, in the automatic steering system for a ship, the automatic steering system comprises a needle holding controller having an integral characteristic and a needle changing controller provided with a feedforward controller without the integral characteristic. It is characterized by having.

According to the present invention, in the automatic steering device for a ship, the controller for changing needle uses a generalized control target in which the control target and the feedforward controller are combined.

According to the present invention, in the automatic steering system for a ship, the above-mentioned needle keeping controller uses a generalized controlled object which is a combination of the above-mentioned controlled object and a weighting function and integral characteristic.

The following documents will be helpful. For details, refer to such a document. (1) "3rd Maneuverability Symposium" textbook, Japan Society for Shipbuilding Test Tank Committee, 243/279 (1982) (2) "SIC Summer Seminar '92 -Automatic steering system design method based on new control theory-" Textbook, Japan Society of Instrument and Control Engineers, 59/60, 1992, (3) "A method of constructing a secondary stabilization compensator for systems with both structural and unstructural uncertainties", Ryu Shintetsu, Koshi Tamura , Institute of Instrument and Control Engineers, vol. 29, no. 1
0, 1171/1175 (1993)

Further, as an example in which the problem of the automatic steering system is reduced to the H∞ control problem and solved, Japanese Patent Application No. 6-1 filed on July 29, 1994 by the same applicant as the present applicant.
No. 78652 is disclosed. See the same application for details.

[0036]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to FIGS. 1 to 5, corresponding parts in FIGS. 6 to 7 are designated by the same reference numerals, and detailed description thereof will be omitted.

FIG. 1 is a block diagram of a control system, that is, an automatic steering system of an automatic steering system for a ship according to the present invention. The automatic steering system of this example includes a first adder 11, an automatic steering device, that is, an autopilot 12, a second adder 13, a controlled object 14, and a third.
And an adder 15 of. The controlled object 14 is the hull 14-1 and the heading detector 14-2 shown in FIG. 6 integrated. The steering gear 16 is omitted here because the response of the steering gear 16 is about one digit faster than the response of the automatic steering system, and the dynamic characteristics thereof can be ignored.

Set course φ _C as an input signal to the automatic steering system
Is entered. The first adder 11 obtains the deviation ERR = φ _C −φ between the set course φ _C and the heading φ. The automatic steering device, that is, the autopilot 12 inputs the deviation ERR and outputs the command steering angle U _C. The command rudder angle U _C is supplied to the second adder 13.

The second adder 13 has a command steering angle U_CMore angle
The converted disturbance d is subtracted, and the result U_C-D is the control target
It outputs to 14. The third adder 15 outputs the controlled object 14.
Force signal φ₀Disturbance d detected by the heading detector 14-2_V
To subtract the heading φ = φ ₀-D_VGenerate bow
Disturbance d detected by the azimuth detector 14-2_VIs the block diagram of Figure 6.
It is not shown in, and is newly added here.

The heading φ is fed back to the first adder 11. The deviation ER which is the output of the first adder 11
A stable closed loop is constructed so that R is zero. At this time, the heading φ is output as an output signal of the automatic steering system.

According to the present invention, for the automatic steering system, H∞
Quadratic stabilization theory based on control theory
c Stabilization) is used. The quadratic stabilization theory deals with the uncertainty of ship parameters by the amount of fluctuation of uncertainty.

The automatic steering device 12 is of the feedback controller type for holding the needle which includes the integrator as described with reference to FIG. 7 and for the variable needle which does not include the integrator as described with reference to FIG. There is a feedback controller type. The method of the invention applies to closed loops with either controller.

According to the H∞ control theory, the automatic steering device 12
Is designed to satisfy the specifications of a feedback controller having a function that combines optimal control and Kalman filter (state observer). As will be described later, controlled object 1
Reference numeral 4 is a combination of a hull model and a heading detector (gyro compass or magnet compass). The hull model has uncertain parameter errors due to model errors due to linearization, the effects of sea conditions, changes in cargo, effects of ship speed, etc.

The H ∞ control indirectly treats this uncertain parameter error as unstructured variation in the frequency domain.
Second-order stabilization theory directly treats this uncertain parameter error as structural uncertainty in the time domain. Therefore, 2
The design based on the secondary stabilization theory naturally fits the controlled object.

The feature of the quadratic stabilization theory is that the uncertainty of the parameter is defined by its fluctuation range, and a generalized controlled object scaled by a constant ε is handled. Also 2
Whether or not next-order stabilization is possible depends on whether or not a Riccati solution that satisfies the Lyapunov inequality can be obtained. Since the solution method of this inequality is equivalent to the solution method of H ∞ control, the method of H ∞ control can be applied to the quadratic stabilization theory as it is.

Further, the optimum setting value of the constant ε is uniquely determined by adding another iterative loop outside the solution loop of the H∞ control. Thus, 2
A next-stabilized control system is required.

In the present invention, as the basic H∞ control,
A method based on the time response characteristic of LQ format was adopted. The reason is that the autopilot has conventionally used a quadratic type evaluation function, and the order of the controller is not increased.

The basic equation of H∞ control will be determined with reference to FIG. As shown in FIG. 2, the H∞ control is a generalized control target 10
It deals with a closed loop including a zero and a compensator or controller 120 ″. As will be explained later, in the case of the present invention, the controller 120 ″.
Is the autopilot 12.

(1) First, the formulas for the evaluation function and the evaluation amount are obtained. For the closed-loop transfer function T _zw (s) from the disturbance input w to the evaluation amount z, the H∞ control norm is defined by the following equation.

[0050]

[Equation 2] ‖z‖ ₂ ≦ ‖T _zw ‖∞ ・ ‖w‖ ₂

[0051] In this case, the left-hand side of ‖-‖ ₂ L ₂ [0, ∞] norm, is the right-hand side of ‖ · ‖∞ show the H∞ control norm. The H∞ control norm limit is defined by defining the H∞ control norm of the transfer function T _zw (s) by a positive constant γ. This is expressed as:

[0052]

[Equation 3] ‖T _zw ‖∞ ≦ γ

As described above, the present invention adopts the H∞ control method based on the time response characteristic. This is optimal control that is related to the LQ-type optimal control problem and minimizes the LQ-type evaluation function for the disturbance w.

[0054]

(Equation 4)

Here, the subscript (·) ^T indicates a transposed matrix. On the other hand, the evaluation amount z is determined by the following equation.

[0056]

Equation 5] _{z = C 1 x + D 12} u z T z = x T Qx + u T Ru

Here, the coefficients C ₁ and D ₁₂ are expressed by the following equations.

[0058]

(Equation 6)

Q and R are semi-definite and positive definite weights, respectively, and Q ^1/2 and R ^1/2 are matrices satisfying the following equation.

[0060]

(7) Q = (Q ^1/2 ) ^T Q ^1/2 , R = (R ^1/2 ) ^T R ^1/2

Therefore, the controller 120 "by the H∞ control
Is for minimizing the evaluation amount z with respect to the disturbance w.

(2) Next, the formula of the generalized controlled object 100 including the uncertainty is obtained. Here, the uncertainty of the parameter is defined by the fluctuation range of the uncertainty, and the generalized controlled object 100 including the weighting function and the controlled object is represented by the following expression.

[0063]

(Equation 8)

Here, x is a state quantity, u is an operation quantity, y is a detected quantity, z is an evaluation quantity, and w is a disturbance. A, B ₁ , B ₂ ,
C ₁ , C ₂ , D ₁₂ , and D ₂₁ are system matrices. These variables and matrices have appropriate dimensions. In addition, t is time, ΔA (t) and ΔB (t) are plant and input fluctuation width matrices, and H, F (t), E ₁ and E ₂ are appropriate matrices. I is an appropriate unit matrix, and the matrix with t is time-varying.

Generalized controlled object 10 including this uncertainty
The H∞ control cannot be applied to 0. Instead, the generalized controlled object 100 that is scaled using the constant ε is used. This is expressed by the following equation.

[0066]

[Equation 9]

[0067]

[0068]

(Equation 10)

Here, x _K is the state quantity, A _K , B _K , C _K
Is a system matrix and variables and matrices have appropriate dimensions. Substituting the controller represented by the equation (10) into the two generalized control targets (equation (8) and equation (9)), the following closed loop systems are obtained.

[0070]

[Equation 11]

Here,

[0072]

(Equation 12)

Whether or not quadratic stabilization is possible for a closed loop system including uncertainty depends on whether or not there is a positive definite solution of the symmetric matrix P _C that satisfies the following Riccati inequality.

[0074]

(Equation 13)

In examining the Riccati inequality of two closed-loop systems derived from two generalized controlled objects, the following relation is used.

[0076]

[Equation 14]

Therefore, the following relational expression is obtained.

[0078]

(Equation 15)

From this relational expression, it can be seen that the two Riccati inequalities shown in the equation 13 have the same value. Therefore, the two generalized control targets represented by the equations (8) and (9) have the same value. Therefore, in the H ∞ control problem, the scaled generalized control target represented by the equation 9 may be used instead of the generalized control target including the uncertainty represented by the equation 8.

(3) Next, the optimum set value of the constant ε is obtained.
When using a scaled generalized control target, it is necessary to set a constant ε. First, the following equation is defined.

[0081]

(Equation 16)

If Z (γ) <0, then H ∞ of the closed loop system
The control norm limit is expressed as:

[0083]

[Equation 17]

This means that H∞ control is possible. Here, the exponent (·) ⁻¹ indicates an inverse matrix. The possibility of second-order stabilization means that H∞ control is possible. A quadratic equation for the constant ε is obtained from the second equation of the equation (13).

[0085]

(Equation 18)

Since the left side of this equation is a quadratic function of the constant ε, it has an extreme value. Therefore, considering that ε> 0 and Z (γ) <0, the following equation holds.

[0087]

Γ (ε) = f {(ε−ε ₀ ) ² } + γ ₀

Here, f {•} indicates a function. This is ε
= Ε ₀ , the constant γ has a minimum value γ ₀ . Therefore, the set value of the constant ε is uniquely determined, and then the value γ of the H∞ control norm limit becomes the minimum.

Next, the configuration of the controller 120 "shown in the equation (10) will be described. The symmetric matrix X, which satisfies the following three conditions:
If Y exists, the controlled object 100 including the uncertainty is 2
It can be stabilized next. Reference is made to the above-mentioned document (3).

(1) The Riccati inequality has the quasi-positive definite solution X ≧ 0.

[0091]

(Equation 20)

(2) The Riccati inequality has a quasi-positive definite solution Y ≧ 0.

[0093]

(Equation 21)

(3) The spectrum radius is 1 or less.

[0095]

[Equation 22] λ _max (XY) <1

These three conditions are the existence conditions of the solution of H∞ control. At this time, the controller 120 ″ is given as follows.

[0097]

(Equation 23)

Here, there is the following relationship.

[0099]

(Equation 24)

Calculation of the control gain of the controller 120 ″ will be described with reference to FIG. 3. FIG. 3 shows a flow of calculation of the controller 120 ″. The controller 120 "calculates the control gain by a flow including the following steps.

101 A control object including uncertainty is represented by a state space expression. 102 Design specifications are defined by an evaluation amount z or a weighting function. 103 A generalized controlled object having uncertainty is constructed using the controlled object and the evaluation amount z or the weighting function.

104 A generalized controlled object having uncertainty is converted into a scaled generalized controlled object. 105 The set values of the weights Q and R of the evaluation amount z are given. 106 The initial value of the constant ε is given. Solving for the constant γ of 107 H ∞ control.

It is determined whether 108 γ is the minimum value. (a) When it is not the minimum value, the constant ε is updated and the constant γ is solved. (b) Hold the constant ε when it is the minimum value. 110 Calculate the controller gain using a predetermined constant.

The optimum set value of the constant ε is obtained by adding an iterative loop of the constant ε to the solution calculation method of the constant γ of the H∞ control as shown in FIG.

Description will be made with reference to FIG. FIG. 4 is a block diagram of an automatic steering system including the autopilot 12 and the controlled object 14. Autopilot 12 has two adders 1
2-1, 12-4 and feedforward controller 12-2
And a feedback controller 12-3. The autopilot 12 inputs the set course φ _C and heading φ = Yp and outputs the command rudder angle U _C.

The controlled object 14 includes a hull model 14-1 and a heading detector 14-2, which inputs a command rudder angle Uc and outputs a heading Yp. The hull model 14-1 is expressed by a linear equation, and the command steering angle Uc is input to turn angular velocity dφ / d.
Output t.

The azimuth detector 14-2 determines the turning angular velocity dφ / d.
It acts as an integral characteristic of inputting t and outputting the heading φ.

The hull model 14-1 has a parameter error due to a model error due to linearization or a reduction in dimension, an influence of a fluid force due to a sea condition, and changes in cargo, fuel and ship speed. It is assumed that the scale factor, bias, and acceleration errors of the azimuth detector 14-2 can be ignored.

The controlled object 14 is represented by the state space expression as in the following equation.

[0110]

(Equation 25)

Here, xp is the state quantity, Uc is the operation quantity (command steering angle), yp is the detection quantity (heading direction φ), w ₁ is the plant disturbance, w ₂ is the detection disturbance, Ap, Bp, Cp, Dp. Is a system matrix to be controlled, and ΔAp and ΔBp are plant and input uncertainty matrices, respectively.

[0112]

(Equation 26)

Here, φ is the heading of the ship, ap and bp are elements of the nominal value (set value), and are composed of the tracking stability index Ts of the hull characteristics and the turning force index Ks. The tracking stability index and the turning force index are collectively referred to as a maneuverability index. Δap, Δ
bp is an uncertainty parameter and is determined using the fluctuation range δ.

[0114]

[Equation 27]

Next, the generalized control object for changing the needle will be described. At the time of changing the needle, the feedforward controller 12-2 works to improve the response characteristic. The feedforward controller 12-2 has the inverse characteristic of the nominal model of the controlled object 14. Consider a closed loop system that includes this. The specification of the needle changing controller does not have the integral characteristic because the response characteristic is more important than the disturbance attenuation characteristic. The needle changing controller 12-2 is shown by the following equation.

[0116]

[Equation 28]

Here, x _KC is the state quantity, U _FB is the feedback steering angle, and e is the deviation e = the set course φ _C and the heading yp.
φ _C −yp. A _KC , B _KC , and C _KC are change needle controllers 1
2-2 is a system matrix. The nominal system is shown by the following equation.

[0118]

(Equation 29)

[0119]

[Equation 30]

Here, the command steering angle U_CIs a feed forwarder
Steering angle U_FFAnd feedback rudder angle U _FBIs the sum of Immediately
Chi, U_C= U_FF+ U_FBIt is. In addition, there is the following relationship.

[0121]

(Equation 31)

Therefore, by performing the following replacement, a generalized controlled object including the uncertainty for changing the needle can be obtained.

[0123]

(Equation 32)

[0124]

[Equation 33]

[0125]

(Equation 34)

Here, qc ₁ , qc ₂ and γc are weight constants for changing the needle.

Next, with reference to FIG. 5, a generalized controlled object for holding the needle will be described. At the time of holding the needle, the set course φ _C is in the zero state, and the feedforward controller 12-2 does not work. The specifications of the needle-holding controller place more importance on the disturbance attenuation characteristic than the response characteristic. Therefore, it has an integral characteristic.

In the controlled object 14, the plant matrix A _p
And the disturbance matrix D _p have no controllability, the weighting function W
The position of the disturbance w ₁ is changed using ₁ (s) to ensure controllability. FIG. 5A is a state before changing the position of the disturbance w ₁ , FIG. 5B
FIG. 4 is a block diagram showing a state after the position of the disturbance w ₁ is changed. The controlled object 14 after the position change shown in FIG. 5B is shown below.

[0129]

(Equation 35)

Here, assuming that the weighting function is a first-order low-pass filter of the time constant T _W1 , the element of the disturbance matrix D _P is cp = 1 / T _W1 .

Next, a generalized control target of the needle holding controller, that is, the feedback controller 12-3 having an integrator is determined. This uses the method of the above-mentioned literature (2).

[0132]

[Equation 36]

Here, Q _K and R _K are weight matrices for needle holding,
Q _I is an integral weighting constant, and D _K and N _K are needle holding weighting constants.

[0134]

(37)

Here, q _K1 , q _K3, and γ _K are needle holding weight constants, and q _I is an integral weight constant. The controller has an integrator by performing the following operation on the controller obtained by this generalized control target.

[0136]

(38)

Here, K _K (s) and K _K ′ (s) are the transfer function of the controller calculated by the generalized controlled object and the transfer function of the controller provided with the integrator, respectively. The fractional term is derived from the term newly added to the generalized control target.

Consider the transfer function K _K ′ (s).
Since the denominator of the transfer function K _K (s) on the right side actually includes the s + 1 term (not described), the s + 1 term in the numerator of the fraction is offset. Therefore, only the s term of the denominator of the fraction remains on the right side. Therefore, an integrator is added to the controller represented by the transfer function K _K ′ (s).
The integrator allows the heading error due to the bias component of the disturbance to be removed.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to these examples, and various modifications can be made within the scope of the invention described in the claims. It will be understood by those skilled in the art.

[0140]

According to the present invention, there is an advantage that the performance of the automatic steering system can be clearly introduced as a design specification.

According to the present invention, since it is possible to design a controller capable of handling a change in the characteristics of a ship as a fluctuation amount of uncertainty, there is an advantage that the stability of an automatic steering system can be guaranteed.

According to the present invention, there is an advantage that a controller having desired characteristics and performance can be constructed by appropriately adjusting the evaluation amount and the weighting function.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration example of an automatic steering system for a ship according to the present invention.

FIG. 2 is a block diagram showing a closed loop subject to the H ∞ control problem which is the basis of the present invention.

FIG. 3 is a flow chart for calculating a controller gain.

FIG. 4 is a block diagram showing a configuration example of an automatic steering system for a ship according to the present invention, which results in an H∞ control problem.

FIG. 5 is a block diagram showing an example in which the position of the disturbance applied to the controlled object at the time of holding the needle is changed.

FIG. 6 is a block diagram showing a configuration example of a conventional marine vessel automatic steering system (for needle change mode).

FIG. 7 is a block diagram showing a configuration example of a conventional marine vessel automatic steering device (for the needle keeping mode).

FIG. 8 is a diagram showing a needle change response characteristic of a conventional marine vessel automatic steering system.

[Explanation of symbols]

 11 Adder 12 Automatic steering device 12-1 Adder 12-2 Feedforward controller 12-3 Feedback controller 12-4 Adder 13 Adder 14 Control target 14-1 Hull, Hull model 14-2 Direction detector 14 -3, 14-4 Adder 15 Adder 16 Steering machine 100 Generalized control target 120, 120 'Automatic steering device 120 "Controller

Claims (5)

[Claims] 1. An automatic steering device for a marine vessel configured to be closed-loop controlled by an automatic steering device which inputs a signal indicating a deviation of a bow direction with respect to a set course and outputs a commanded steering angle. Treat the control object consisting of the hull and heading detector as input with the above heading as output and treat it as a generalized control object with uncertainty, and combine the generalized control object with the automatic steering device to reduce to the H∞ control problem. , H of the transfer characteristic from the disturbance of the generalized control target to the evaluation amount
∞ Automatic norm for ships characterized by making the ∞ control norm less than the specification and converting the generalized controlled object having the above uncertainty into a generalized controlled object scaled using a constant ε and resulting in an H∞ control problem. Steering device. 2. The marine automatic steering apparatus according to claim 1, wherein the constant ε minimizes the transfer characteristic of the H∞ control norm from the disturbance to the evaluation amount in the scaled generalized control target. An automatic steering device for ships, which has an optimum set value of 3. The automatic steering apparatus for a ship according to claim 1, wherein the automatic steering apparatus has a needle holding controller having an integral characteristic and a feed forward controller having no integral characteristic. An automatic steering device for ships, comprising: a controller. 4. The automatic steering system for a marine vessel according to claim 3, wherein the controller for changing needles uses a generalized controlled system in which the controlled system and the feedforward controller are combined. apparatus. 5. The marine vessel automatic steering apparatus according to claim 3 or 4, wherein the needle keeping controller uses a generalized controlled object that combines the controlled object with a weighting function and an integral characteristic. Automatic steering device.