JP2003022134A - System and simulator for controlling position of floating body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To establish efficient position control technology for relative motion floating bodies having huge mass to be dynamically influenced with each other. SOLUTION: A system for controlling positions of floating bodies having a first floating body 1 and a second floating body 3 linked with the first floating body 1 to be displaced. The first floating body 1 is provided with a control circuit 6 to control a motion of the first floating body 1. The control circuit 6 is provided with a first control circuit 7 to output a first output signal 9 based on input of a first input signal θ1-θ2 as physical quantity regarding position control of the first floating body 1 and a second control circuit 8 to output a second output signal 11 based on input of second input signals R, ΔR as relative physical quantity of the first floating body 1 to the second floating body 3. The motion of the first floating body 1 is controlled based on the first output signal 9 and the second output signal 11. The second control circuit 8 is a neutral network to output the second output signal 11 to compensate the first output signal 9.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a floating body position control system, and more particularly, to a floating body position control system for performing positional control between two floating bodies that mechanically influence each other.

[0002]

2. Description of the Related Art The ocean has many possibilities as a development site for energy resources such as oil, gas and methane hydrate. Also, as one of the greenhouse gas emission sites, a plan to bury it in a deep seabed is being considered, and it is expected that the use of the ocean (seabed) will continue to increase. Along with this, it is expected that development in the ocean will expand not only to conventional continental shelves but also to deeper waters such as slopes of continental shelves and waters with more severe natural environmental conditions. In this situation, the floating production platform (FPS
The floating production system using O) has been attracting attention as a system suitable for the development of marine resources such as small-scale offshore oil fields, and has a proven track record.

However, as the development of marine resources such as small-scale offshore oil fields progresses to medium and deep water depths, the FPS has traditionally been used.
Only the mooring method used as the O position holding method is difficult from the technical and economical point of view. Therefore, as a floating production system, a DPS (Dynamic Po) using a plurality of thrusters as actuators is used.
Positioning control by a positioning system (a system for returning the position to the original position by controlling the thruster when the FPSO is flown from a predetermined position due to sea conditions and weather) is indispensable. And in this case,
It is necessary to properly control each thruster, which is an actuator, in accordance with changes in the natural environment, and hold the floating production platform in an appropriate position.

Further, in order to improve the operating rate of the entire production system, improving the operating rate at the time of offshore loading from the FPSO to the shuttle tanker is also an important issue.
In order to improve the operating rate at the time of offshore shipping, it is necessary to suppress the relative movement between the FPSO and the shuttle tanker to raise the limit wave height at the time of shipping. The greatest problem in raising the limit wave height is the hawser tension between the FPSO and the two floating bodies of the shuttle tanker.

That is, it is required to optimize the position control between the two floating bodies. The offshore production system comprises an FPSO moored offshore and whose position is controlled approximately constant.
Offshore loading is performed from the FPSO to the shuttle tanker. Relative position control between the FPSO and the shuttle tanker is desired. Conventionally, the FPSO and the shuttle tanker automatically adjust the relative distance and the relative rotation angle by entrusting themselves to the tidal current, or by using the thruster of the tugboat, the FPSO and the shuttle tanker, artificially and artificially control the control. I was doing.

Such position control of FPSOs and shuttle tankers is desired to be automated. As shown in FIG. 1, the shuttle tanker 3 is connected to the FPSO 1 by a hawser 4, but is left in a natural discharge state in a natural environment such as tidal currents and waves. Shuttle tanker 3 and FP
The distance between the SO1 reference points is maintained at a constant distance within the range allowed by the sagging and expansion / contraction of the hawser 4. Shuttle tanker 3 rotates FPS due to change in direction of external force
FPSO when moving / displacement relative to O1
It is desired that the feedback control be performed so that the relative distance between 1 and the shuttle tanker 3 becomes constant within the allowable range.

As such control, PID control is appropriate. As such PID control, a constant distance control that controls the rotational displacement of the FPSO 1 so that the distance between the reference points is constant can be considered. The movable items of FPSO are
Since it is limited to the turning angle, in the constant distance control, the Y-axis (horizontal axis, axis orthogonal to the stern bow direction) direction component of the relative position of the object fixed coordinate system is converted into a deviation of the turning angle. It was observed that the constant distance control reduces the tension on average, but since there is no part that stabilizes the turning angle, the deflection becomes large and the tension becomes excessive.

In addition, the PID control is based on F with a huge mass.
In some situations, it is difficult to satisfactorily address the non-linear phenomenon in which the disturbance response of the PSO is not rapid and the relative distance displacement immediately affects the tension of the hawser 4. For example, this is the case when the external force (disturbance) of the sea condition / weather such as high wave height is large. In such a case, the relative rotation of the two floating bodies immediately causes a distance variation between both ends of the hawser, the correspondence between the target control amount and the output control amount is delayed, and the physical quantity between the various two floating body relatives is delayed during the delay. It is difficult to realize PID control alone to satisfactorily perform a relative motion floating body having a huge mass due to non-linear factors that are compounded such that the control fluctuates uncontrollably and the control is a multibody problem. Is.

It is required to establish a technique for effective motion control of a relative motion floating body having a huge mass that mechanically influences each other. In particular, effective control of the movement between two moving floating bodies whose rotary movement causes relative distance fluctuations is desired. In particular, it is required to establish a correction technique for correcting the rotational movement based on the relative distance and its variation, and it is desirable that the legality against disturbance is excellent.

[0010]

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to effectively control the relative position between two floating bodies having huge masses that mechanically influence each other, such as an offshore FPSO and a shuttle tanker. It is to provide a possible floating body position control system.

Another object of the present invention is to provide an offshore FPS.
An object of the present invention is to provide a floating body position control system capable of effectively controlling the relative distance between two floating bodies having enormous masses that mechanically influence each other, such as O and a shuttle tanker, even in the presence of an external force.

Still another object of the present invention is to provide offshore FPS.
An object of the present invention is to provide a floating body position control system such as an O and a shuttle tanker, which is capable of effectively controlling the relative distance between two floating bodies in which rotational movement causes fluctuations in the relative distance.

Still another object of the present invention is to provide an offshore FPS.
It is an object of the present invention to provide a floating body position control system capable of effectively controlling the tension applied to the hawser connecting the O and the shuttle tanker to a low level.

Still another object of the present invention is to provide offshore FPS.
It is an object of the present invention to provide a floating body position control system that performs control capable of raising the limit wave height during the offshore shipping of oil performed by O and a shuttle tanker.

Another object of the present invention is to provide an offshore FPS.
An object of the present invention is to provide a floating body position control simulator and a floating body position control simulation method capable of simulating the positional control between two floating bodies having huge masses that mechanically influence each other, such as O and a shuttle tanker.

[0016]

[Means for Solving the Problems] Means for solving the problems will be described below by using the numbers and symbols used in the embodiments of the present invention. These numbers and signs are added to clarify the correspondence between the description of [Claims] and the [Embodiment of the Invention]. However, those numbers and signs should not be used for the interpretation of the technical scope of the invention described in [Claims].

In order to solve the above problems, the floating body position control system of the present invention comprises a first floating body (1) and a second floating body (3) displaceably connected to the first floating body (1). It is equipped with. Then, the first floating body (1) is controlled based on the direction (θ1) of the first floating body (1) and the direction (θ2) of the external force acting on the first floating body. A PID control circuit (7) for controlling the movement of the first floating body (1) is provided so as to direct the external force in the direction (θ2).

Further, the floating body position control system of the present invention is
Among the gains of the PID control circuit (7), the proportional gain is adjusted so as to show a tendency to move the first floating body in the direction of the external force, the integral gain is adjusted to substantially zero, and the differential gain is The first floating body is adjusted so as not to whirl.

Further, the floating body position control system of the present invention is
It comprises a first floating body (1) and a second floating body (3) displaceably connected to the first floating body (1). The first floating body (1) includes a control circuit (6) for controlling the movement of the first floating body (1), and the control circuit (6) controls the position of the first floating body (1). The first physical quantity involved
A first control circuit (7) that outputs a first output signal (9) based on the input of the input signal (θ2-θ1), and the relative position of the first floating body (1) with respect to the second floating body (3). Based on the input of the second input signal (R, ΔR) that is a physical quantity, the second
A second control circuit (8) for outputting an output signal (11). The movement of the first floating body (1) is controlled based on the first output signal (9) and the second output signal (11), and the second control circuit (8) controls the first output signal (9). The neural network outputs the second output signal (11) for correcting (9).

Further, the floating body position control system of the present invention is
The first signal (9) acts on the first floating body (1)
The direction (θ1) of the force with respect to the first floating body,
Two signals (11) are the first floating body (1) and the second floating body.
Relative distance between (3) (Y _len) Based on the amount and
Displacement of relative distance (ΔY_len) And the quantity based on.

Further, the floating body position control system of the present invention is
It comprises a first floating body (1 ') and a second floating body (3) displaceably connected to the first floating body (1'). The first floating body (1 ′) controls the movement of the first floating body (1 ′) based on a first input signal which is a physical quantity related to the position control of the first floating body (1 ′). 1 control circuit (8 '), the said 1st control circuit (8') is a neural network (FIG.3 (a)).

Further, the floating body position control system of the present invention is
It comprises a first floating body (1 ') and a second floating body (3) displaceably connected to the first floating body (1'). The first floating body (1 ') includes a control circuit for controlling the movement of the first floating body (1'), and the control circuit is a physical quantity related to the position control of the first floating body (1 '). There is a first
A first output signal that outputs a first output signal based on an input of the input signal
A control circuit (1 ′) and a second control circuit (8) that outputs a second output signal based on the input of a second input signal that is the first output signal and the motion output of the first floating body (1 ′). ') And. The movement of the first floating body (1 ') is controlled based on the first output signal, and the second control circuit (8').
Is a neural network that outputs the second output signal for correcting the output signal of the first control circuit (FIG. 3B).

Further, the floating body position control system of the present invention is
The first signal is a direction of an external force acting on the first floating body (1, 1 ').

Further, the floating body position control system of the present invention is
The motion output of the first floating body (1, 1 ') is returned to the neural network (8, 8') as teacher data,
The neural network (8, 8 ') normally performs learning based on the teacher data.

Furthermore, the floating body position control system of the present invention is
The neural network (8, 8 ') controls the first floating body (1, 1') after learning is performed in advance.

Further, the floating body position control system of the present invention is
The first control circuit (7, 7 ') is a PID control circuit.

Furthermore, the floating body position control system of the present invention is
The second floating body (3) is connected to the first floating body (1, 1 ') via a hawser (4).

Further, the floating body position control system of the present invention is
The neural network (8, 8 ') is a multi-layer (23
-1 to 3). Then, the multilayer (23-1 to 23-1
3) the i-th neuron output of one layer is Q
_The neuron input represented by i and input to the j-th neuron in the layer next to the one layer is represented by Q _j , and the weighting factor between the i-th neuron and the j-th neuron Is represented by ω _ij , and the neural network (8, 8 ′) has the following formula: Q _j = f (Σ _i ω _ij Q _i ).
It is represented by.

Further, the floating body position control system of the present invention is
The formula is Q_j= Tanh (Σ_iω _ijQ_i)
It

In order to solve the above problems, the floating body position control simulator of the present invention comprises an actuator simulation input circuit (inside 16) to which an operation signal for operating the actuator of the first floating body is input, and the movement of the first floating body. A floating body simulation output circuit (inside 16) for outputting a motion value formed from the first motion value of the first floating body and the second motion value of the movement of the second floating body displaceably connected to the first floating body; A neural network (inside 17) is provided between the actuator simulation input circuit and the floating body simulation output circuit that receives a signal and outputs the motion value.

Further, in the floating body position control simulator of the present invention, the neural network (8, 8 ') includes multiple layers (23-1 to 3). Then, the multilayer (23-
The output of the i-th neuron of one of the layers 1 to 3) is represented by Q _i , and the input of the neuron input to the j-th neuron of the next layer of the one layer is represented by Q _j , The weighting factor between the i-th neuron and the j-th neuron is represented by ω _ij , and the neural network (8, 8 ′) has the following formula: Q _j = tanh (Σ _i ω
_ij Q _i ).

In order to solve the above-mentioned problems, the floating body position control simulation method of the present invention comprises a step of generating a first motion equation which is a motion equation of the first floating body, and the first motion equation.
Calculating a first external force, which is the sum of the forces exerted on the floating body, from the first equation of motion and the first external force,
A step of solving the first equation of motion; a step of generating a second equation of motion that is a equation of motion of a second floating body that is displaceably connected to the first floating body; and a sum of forces exerted on the second floating body. Of the second external force, the step of solving the second equation of motion from the second equation of motion and the second external force, and the connecting hawser connecting the first floating body and the second floating body. Calculating the tension.

In order to solve the above problems, the program of the present invention is a program for executing the above-mentioned floating body position control simulation method.

[0034]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of a floating body position control system according to the present invention will be described below with reference to the accompanying drawings. In the present embodiment, the offshore FP is used as the two floating bodies.
The SO and the shuttle tanker will be described as an example. However, the floating body position control system according to the present invention can be applied to other two floating bodies having huge masses that influence each other mechanically.

FIG. 1 shows an FPSO 1 and a shuttle tanker 3 and their peripheral equipment constituting a two-floating body to which the floating body position control system of the present invention is applied. It has FPSO 1, mooring line 2, shuttle tanker 3, hawser 4, and floating hose 5.

In one embodiment of the present invention, a floating body position control system capable of effectively controlling the position of the two floating bodies having the structure as shown in FIG. 1 in the presence of an external force, and the floating body position thereof. Provided is a floating body position control simulator capable of simulating control.

The FPSO 1 as the first floating body is moored by a plurality of mooring lines 2 in the upper sea area of the ocean oil field and is almost fixed on the ocean, but it can be slightly displaced on the two-dimensional plane. , Can be rotated on its two-dimensional plane. The FPSO1 generally rotates about a vertical line L passing through the center of gravity G. The center of gravity G can swing in a narrowly restricted and closed area. It is a development base for offshore oil fields.
Performs drilling work in oil fields, crude oil removal work, crude oil storage work and related work. Then, the crude oil taken out from the oil field and stored inside is sent to the shuttle tanker 3.

A thruster, which is an actuator for propulsion and position holding, is provided below the water surface. Then, by controlling the thruster based on a control algorithm to be described later, proper position holding is performed in the situation where the shuttle tanker 3 is connected. It also has measuring devices (not shown) for operating conditions and positions (including shuttle tankers), sea conditions, and weather.

The mooring cable 2 is a cable or cable for mooring for anchoring the FPSO 1 in a substantially specific range. The mooring line 2 is attached, operated and controlled by a windlass, a winch, a mooring post, a mooring hole and other accessories for mooring.

The shuttle tanker 3 as the second floating body,
It is connected by the connecting hawser 4 and left and discharged in the natural state on the downstream side of the tidal current. Almost constrained by the length of the connecting hawser 4, the connecting reference point of the FPSO 1 (for example, FPS
The distance between the stern of O1) and the connection reference point of the shuttle tanker 3 (for example, the stern of the shuttle tanker 3) is naturally maintained at a substantially constant value, but the extent of expansion and contraction depending on the tension fluctuation of the connection hawser 4 is not sufficient. The relative distance varies within the range.
Liquids such as crude oil and natural gas stored in the FPSO 1 are loaded using the floating hose 5. Unloaded crude oil is transferred to other large tankers and tanks (combinates) on land. When connected to FPSO1, it does not perform operation control by itself.

The connecting hawser 4 as the hawser is an FPS.
A cable or cord that connects the O1 and the shuttle tanker 3. Connection part 100-1 between FPSO 1 and hawser 4, and connection part 1 between shuttle tanker 3 and connecting hawser 4.
00-2 is a hydraulic device (described later). And
By controlling the hydraulic pressure, in addition to adjusting the tension applied to the hawser 4, the hawser 4 for coupling is protected from sudden fluctuations.
Does not easily get damaged.

With reference to FIG. 25, the connecting section 100-1 (similarly to the connecting section 100-2) will be described. FPSO1
The connecting hawser 4 connecting the shuttle tanker 3 and the shuttle tanker 3 is FP
A hydraulic device 102 is connected to the SO1 and the shuttle tanker 3. The hydraulic device 102 has a measuring unit (not shown) that measures the tension applied to the linkage hawser 4, and transmits the signal to the hydraulic control unit 101. When a sudden increase in tension occurs, the hydraulic control unit 101 outputs a signal for adjusting the hydraulic pressure to absorb the impact to the hydraulic device 102. Further, similarly, it is possible to pull or loosen the connecting hawser while keeping the tension constant (or a constant range).

One of the floating hoses 5 is FPS.
A hose for transporting crude oil connected to O1 and the other to the shuttle tanker 3. It is made longer than the Hawser 4 so that unnecessary force such as tension is not applied. Since the specific gravity of the material is 1 or less and the specific gravity of crude oil is 1 or less, it floats on the sea surface.

First, a mathematical model used for relative position control in the floating body position control system and the floating body position control simulator according to the present invention will be described. (D) In floating body relative position control, the positions of the two floating bodies are calculated based on this mathematical model. Then, various controls are performed based on the result.

FIG. 2 shows a coordinate system which is a prerequisite for establishing the equations of motion of the FPSO 1 and the shuttle tanker 3.
The horizontal axis is the y ₀ axis and the vertical axis is the x ₀ axis. The symbols in the figure are as follows. [delta] _F: FPSO Thruster swing angle, _n S-F: FPSO thruster propeller speed, [psi _F: FPSO azimuth, _r F: FPSO azimuth angular velocity, phi _F: FPSO roll angle, theta _F: FPSO pitch angles, X _0-M : FPS OX direction mooring position, Y _0-M : FPS YO direction mooring position, F _X : FPSO turret X direction mooring force, F _Y : FPSO turret Y direction mooring force, T _S-F : FPSO thruster thrust, T _{H-F = T H-S} : FPSO-S / T Hosa tension, ψ _H-F: FPSO side Hosa tension relative angle, _{H W:} wave height, δ _{S: S} / T steering angle, _{n P-S:} S / T propeller rotation speed, n _S-B : S / T Bowus raster rotation speed, ψ _S : S / T azimuth angle, r _S : S / T azimuth angular velocity, φ _S : S / T roll angle, θ _S : S / T pitch angle, F _N : S / T rudder direct pressure, F _T : S / T rudder parallel force, T _P : S / T propeller thrust, T _S-B : S / T bow thruster thrust, ψ _TUG : S / T tag tension direction ψ _H-S : S / T side hawser tension relative angle (S / T: shuttle tanker)

Basic Equation of Motion First, an equation of motion, which is the basis of the movement, is obtained. FPSO
For each floating body (platform) of the shuttle tanker 1 and the shuttle tanker 3, the equations of motion in the coordinate system shown in FIG. 2 are as follows: the floating body's front-back direction (Surge: surge), lateral direction (Sway: sway), and bow sway (turning) direction ( Yaw: yaw) is expressed as follows.

[Equation 1] The meaning of each code is as follows. m: floating mass, _m x: additional mass in the x-axis direction, _m y: additional mass in the y-axis direction, I _zz: floating about the center of gravity of the moment of inertia _{J zz:} additional moment of inertia about the floating body centroid u: object fixed coordinate floating motion of the x-axis direction component v in the system: y-axis direction component of the floating body motion in the object fixed coordinate system r: rate of turn _{X H, Y H, N H} : fluid force floating body is subjected (the longitudinal force, lateral force, moment) _{X P, Y P, N P} : fluid force propeller occurs (longitudinal force, lateral force, _{moment) X T, Y T, N} T: fluid force generated by the thruster (longitudinal force, lateral force, moment) _{X R} , _Y R, _{N R:} fluid force generated by the steering (longitudinal force,
Lateral force, _{moment) X A, Y A, N} A: the external force due to wind (longitudinal force, lateral force, _{moment) X W, Y W, N} W: external force from the wave (longitudinal force, lateral force, moment) _{X HS,} Y _{HS, N HS:} tension by connecting Hosa (longitudinal force, lateral force, _{moment) X M, Y M, N} M: mooring force by the mooring (longitudinal force, lateral force,
Moment) (However, longitudinal force, lateral force, and moment correspond to surge, sway, and yaw directions.) Each external force shown on the right side of (1) to (3) in Equation 1 will be described below. .

Fluid received by the floating body (platform)
Power: X_H, Y_H, N_H Fluid that acts on both FPSO1 and shuttle tanker 3 in common
The force (mainly the fluid force due to the tidal current) is a function of u, v, and r
Is expressed as follows. That is, there is a forward speed
In this case, according to the expression of maneuverability of the ship, (1)-
X in (3)_H, Y _H, N_HCan be expressed as

[Equation 2]

[Equation 3]

[Equation 4] The meaning of each code is as follows. [rho: density of water L: Platform representative length _{C XC, C YC, C NC} : tide force coefficient _{u C, v} C: to water velocity _{U C:} to water velocity ^{_{(U C 2 = u C 2}} + v C 2 ) D: draft β: lateral flow angle μ _C : tidal current direction X ′ _vv , X ′ _rr , X ′ _vr , X ′ _uu : longitudinal fluid force coefficients Y ′ _v , Y ′ _r , Y ′ _vv , Y ′ _rr , Y _'v r: left-right direction of the fluid force coefficient _{N' v, N 'r,} N' vv, N 'rr, N' vr: fluid force coefficient of stem turning motion

Here, the following relationship holds between the water velocity and the tidal current velocity V _C.

[Equation 5]

Fluid generated by propeller of main propulsion device
Power: X_P, Y_P, N_P (A) FPSO1 Since the FPSO1 is not equipped with a main propulsion device,
The lopera force was set to zero as follows. That is, the number 1
X in (1) to (3)_P, Y_P, N_PCan be expressed as
It

[Equation 6] (B) Shuttle tanker 3 It is assumed that the shuttle tanker 3 is equipped with a normal propulsion device and is modeled as follows. That is, _{X P,} Y _{P, N} P number 1 (1) to (3), expressed as follows.

[Equation 7] The meaning of each code is as follows. T _P : Propeller thrust t: Thrust reduction rate

Hydrodynamic force generated by thruster: X_T,
Y_T, N_T (A) FPSO1 The FPSO1 is equipped with a swing-type thruster.
And model it as follows. That is, the number 1 (1)
~ (3) X_T, Y_T, N_TCan be expressed as

[Equation 8] The meaning of each code is as follows. T _S-F: FPSO swing thruster thrust [delta] _F of: FPSO the swing thruster swing angle _x S-F: The swing thruster mounting position x-coordinate (b) shuttle tanker 3 shuttle tanker 3 FPSO, tunnel It is modeled as follows, assuming that the thruster of will be maintained. That is, X _T , Y _T , and N _T in (1) to (3) of Formula 1 can be expressed as follows.

[Equation 9] The meaning of each code is as follows. T _S-B: the bow thruster of the shuttle tanker thrust _{x S-B:} a shuttle tanker bow thruster mounting position x
Coordinate

Fluid force generated by rudder: X_R, Y_R, N_R (A) FPSO1 Since the FPSO1 is not equipped with a rudder,
Treat as zero. That is, (1)-
X in (3)_R, Y_R, N_RCan be expressed as

[Equation 10] (B) Shuttle tanker 3 The fluid force by the rudder provided in the shuttle tanker 3 is expressed as follows. That is, (1) to (3) of Equation 1
X _R , Y _R , and N _R of can be expressed as follows.

[Equation 11] The meaning of each code is as follows. F _N : Rudder direct pressure δ _S : Rudder angle x _R Rudder position x coordinate

External force (sea condition, weather, etc.) model. Disturbances affecting offshore floating platforms
The (external force) is the wind, the wave, and the tidal current.
Fluid force (floating body (platform)
Fluid force on the_H, Y_H, N_HReference)
Next, we will discuss wind, waves, and waves. (A) Wind model (external force by wind): X_A, Y_A, N_A If the external force due to wind can be evaluated with a steady value,
Express like this. That is, in (1) to (3)
X_A, Y_A, N_ACan be expressed as

[Equation 12]

[Equation 13] The meaning of each code is as follows. [rho _A: Density of air _{A AT:} Water line front projection area _{A AL:} Water line side projected area _{U A:} Wind _{V A:} Relative wind velocity _{u A, v} A: x-axis direction component of the relative wind velocity, y-axis direction component u , V: x-axis direction component of platform speed, y-axis direction component β _A : wind direction Ψ: turning angle C _XA , C _YA , C _{NA of} platform: wind pressure coefficient (front-back direction, lateral direction, turning direction) spatially External force due to a fluctuating wind that is uniform and fluctuates only over time can also be considered. The irregular fluctuating wind expresses a designated wind spectrum, and is therefore expressed by superposing the regular fluctuating winds of the elements as shown in the following equation. That is, in Expression 13, Expression 14 is substituted into (32) and (33).

[Equation 14]

(B) Wave model (external force by wave): X_W,
Y_W, N_W The wave model is expressed by the sum of long-period fluctuation drift force and wave forcing force.
Be done. That is, X of (1) to (3) of the equation 1_W= X_WD
+ X_WE, Y_W= Y_WD+ Y_WE, N_W= N_{W D}+ N
_WEIs. And X_WD, Y_WD, N_WDAnd X
_WE, Y_WE, N_{W E}Can be expressed as (B-1) Long-period fluctuation drift force model. According to the method of Pinkster-Newman, the fluctuation
The drift force is calculated as follows using the component waves that make up the irregular wave.
Expressed.

[Equation 15]

[Equation 16]

[Equation 17] The meaning of each code is as follows. (X, Y): Center position of ship weight in fixed space coordinate system a _i : Amplitude of element wave N: Number of element waves g: Gravitational acceleration κ _i : Wave number of element waves ω _i : Circular frequency of element waves ε _i : Element Random initial phase of wave β _W : Wave direction C _XWD , _CYWD , C _NWD : Wave drift force coefficient in regular wave (front-back direction, lateral direction, turning direction) (b-2) Wave force force model Wave force force The mathematical model is expressed by the following formula.

[Equation 18] The meaning of each symbol is as shown below. f _XWE (ω _n ), f _YWE (ω _n ), f
_NWE (ω _n ): Wave force compulsive force amplitude by a unit length amplitude regular wave (front-back direction, lateral direction, turning direction) ε _XWE (ω _n ), ε _YWE (ω _n ), ε
_NWE (ω _n ): Phase of wave forcing force with respect to wave (front-back direction, lateral direction, turning direction)

Hawser force (tension by connecting hawser):
X_HS, Y_HS, N_HS The force of the connecting hawser is FPSO and the shuttle tanker.
In either case, it is expressed as follows. That is,
X of (1) to (3) of number 1_HS, Y_HS, N_{H S}Is next
Can be expressed as

[Formula 19] The meaning of each code is as follows. T _HS : Hawser tension x _HS : Hawser mounting position x coordinate in fixed hull coordinate system Ψ _HS : Hauser direction in fixed hull coordinate system

Mooring force by mooring: X_M, Y_M, N_M The mooring capacity of mooring is limited to the moored FPSO.
And express it as follows. That is, (1)-
X in (3)_M, Y_M, N_MCan be expressed as

[Equation 20] The meaning of each code is as follows. X _M0 , Y _M0 , N _M0 : longitudinal force, lateral force and turning moment at the mooring point x _M : mooring point x-coordinate in the hull fixed system, and X _M0 , Y _M0 , N _M0 are the mooring points of the FPSO. It is expressed as a function of the space fixed coordinate system x _M0 , y _M0 as follows.

[Equation 21]

In the mathematical model, the FPSO 1 and the shuttle tanker 3 are each represented by (1)-(1)
The equation of motion is completed by substituting the external force and the like in Equations 2 to 21 into the equation of motion of (3). The completed equation of motion is
Used to control the relative position of the two floating bodies.

Next, a floating body position control system and a floating body relative position control control method in the floating body position control simulator based on such a mathematical model will be described.

The control method used is PID control (rotational angle azimuth holding control: described later), PID control (rotational angle azimuth holding control: described later) + neural network control (control that constantly learns during control: described later), PID control + There are three methods of neural network control (learning is performed in advance before control, and control is performed using a learning file as a result thereof: described later).

First, the PID control method will be described. FIG. 2 shows a floating body arrangement assumed in the two-floating body relative position control. The FPSO1 is moored, and the only controllable state quantity is the turning angle. In such a case, PID control includes constant distance control that controls the rotational displacement of the FPSO1 so that the distance between the reference points between the two floating bodies (FPSO1 and shuttle tanker 3) is constant, and differential gain-based adjustment (relative Adjustment to reduce the angular velocity) FPSO1
It is conceivable that the rotation angle azimuth holding control is performed by rotating the shaft in the external force direction. However, the constant distance control is not preferable because the tension applied to the hawser (hauser tension) may become excessive.
Therefore, in the present invention, the FPSO 1 performs PID control of moment control (turning angle azimuth holding control in the external force direction). The control purpose is to control the turning angle to reduce the hawser tension between the two floating bodies. Note that the shuttle tanker 3 does not have a control system in order to handle an unspecified shuttle tanker.

In the PID control of the FPSO1 in this embodiment, the FPS is determined from the direction of the external force and the direction of the FPSO1 itself.
This is a motion control for controlling the moment N _T generated by the swinging thruster of O1 to direct the turning angle of FPSO1 in the direction of the external force (control for maintaining the turning angle in the direction of the external force). That is, based on the input of the external force direction and the direction of the FPSO 1 itself to the control device that performs the PID control, the optimum FPS is obtained.
This is a control for outputting the moment N _{T of the} swing thruster of O1.

Then, in the equation (8), N _T output from the PID control device and X _T ² + Y _T ² = T _S-F ²
_(However, X T: FPSO X direction of the fluid force generating the oscillating thruster to _the, Y T: FPSO Y direction of the fluid force generating the oscillating thruster to _{the, T} S -F: FPSO swing thruster thrust) of From the relationship and the minimum control of the thruster thrust T _S-F , T _S-F and δ _F (the swing thruster swing angle of FPSO) are determined. Then, moment control can be performed by controlling T _S-F and δ _F of the swing thrusters of the FPSO1. The minimum control of the thruster thrust T _S-F may be the minimum control of the thruster swing angle δ _F.

In the equation (8), the PID control device
Output from N_TAnd X_T ^Two+ Y _T ^Two= T_SF ^Twoof
Solve the relationship and the equations of motion (1) to (3) in Equation 1.
T_SF, Δ_FIt is also possible to determine motion
The data needed to solve the equation are in FPSO1
Sea conditions, weather, FPSO1 and shuttle tanker driving
Available with various measuring devices and sensors to grasp the situation
Is.

The moment control (turning angle azimuth keeping control in the direction of the external force) realizes stabilization by stabilizing the ship position and reducing the phase shift with respect to the shuttle tanker 3, so that stable control is realized for a long time. You can However, such PID control is not always easy to cope with the non-linear phenomenon in which the disturbance response of the FPSO 1 having a huge mass is not quick and the relative distance displacement immediately affects the tension of the hawser 4. . And
Accordingly, it is difficult to raise the work limit wave height, which is the work limit wave height, in the motion control of the two-floating body coupling on the sea. Therefore, in the present invention, in order to compensate the PID for the behavior caused by a large disturbance, a controller using a neural network having a learning function (hereinafter also referred to as “NN”) is introduced and applied to control.

Control of a non-linear phenomenon under a large disturbance of a huge floating body, which is not always easy only by PID control,
The introduction of a neural network (NN) enables effective implementation.

Here, a control method for performing control using the NN will be described with reference to FIG. In FIG.
The Controller 7'is a P as the first control circuit.
It is a control device such as an ID control unit. NN8 'is the second
A neural network control unit as a control circuit and a neural network. Plant1 '
Is a floating body such as an FPSO or an offshore platform.

FIG. 3A is a control block diagram of a control method called direct type control. Direct NN to Contro
This is the simplest method to use as an ller (controller). In this type, NN is learned so as to be an inverse model of Plant (plant). In this case, PI normally performed by the controller
D control is not performed.

With respect to the input of the target value A0 of the controlled variable A,
NN determines an appropriate value B0 of the control amount B corresponding to the target value A0. Then, B0 is output to Plant. Plan
t changes the control amount B to B0, and the control result A of the controlled amount A
1 is output. The NN uses the comparison result (A0-A1) of the target value A0 of the controlled variable A and the control result A1 as teacher data to perform learning so that A0 and A1 are equal. Then, at the same time as learning, control is continued. When applied to this embodiment, Plant is FPSO1, controlled amount A is the difference between the direction of the external force and the direction of FPSO1 itself, and controlled amount B is the moment generated by the swing thruster of FPSO1.

FIG. 3B is a typical control block diagram of a control method called indirect control. The NN is used as an identifier, etc., and the result shows that the NN is the Controller
Is a way to tune. With this method, the knowledge of conventional control can be used, but the learning effect is only indirectly used.

With respect to the input of the target value A0 of the controlled variable A,
Controller is a control amount B corresponding to the target value A0
The appropriate value B0 of is calculated. Then, B0 is output to Plant. The Plant changes the control amount B to B0 and outputs the control result A1 of the controlled amount A. NN is the controlled variable B0
Based on the input of (and control result A1), the NN outputs an appropriate value A2 of the controlled variable predicted based on the knowledge of the conventional control, and compares the control results A1 and A2. Then, using the comparison result (A2-A1) as teacher data, A2 and A1
Learn so that and become equal. At the same time, A
2 is output to the controller. Control
The ler corrects B0 based on the comparison between the target values A0 and A2. When applied to this embodiment, the Plant is F
PSO1 and the controlled variable A depends on the direction of the external force and the FPSO1.
The control amount B is a difference from the own direction, and is a moment generated by the swing thruster of the FPSO1.

FIG. 3C is a typical control block diagram of a control method called compensation type control. Pl for NN
This is a method used as an ant compensator, and is intended to obtain desired control performance by compensating for complicated dynamics between two floating bodies.

The target value A0 of the controlled variable A is compared with the fed back controlled amount A1. Then, the comparison result (A0-A1) is input to the controller. Controller is the comparison result (A0-A
The appropriate change amount dB of the control amount B corresponding to 1) is obtained. Then, the dB is output to the Plant. The correction value ΔB from the NN is added to dB, and (dB + ΔB) is set to P
Input to lant. Plant sets the control amount B to (dB
+ ΔB), and the control result A2 of the controlled variable A is output. The NN outputs the correction value ΔB of the control amount B based on the input of the change amount ΔA1 of the control result. Along with that
Learning is performed so that ΔA1 becomes 0 by using dB output from the controller as teacher data. When applied to this embodiment, the Plant is FPSO1 and
The controlled variable A is in the direction of the external force, and the controlled variable B is FPSO1.
This is the moment generated by the swing thruster.

FIG. 4 shows a control block diagram of the FPSO1 when the NN control is used as the compensation type in the present embodiment. The control device (not shown) that controls the FPSO1 is
The control computer 6 of the FPSO 1 of FIG. 4 is provided. The control computer 6 includes a PID control unit 7 and a neural network control unit 8. In addition, FPSO1 in FIG.
Is controlled based on the data output by the control computer 6, and outputs the control result. In this embodiment, the swing thruster is controlled based on the data output by the control computer 6 as a control circuit, and as a result, the result that the turning angle of the FPSO 1 is controlled is output.

The correspondence between FIG. 4 and FIG. 3C is PID.
Control unit 7 = Controller (= first control circuit), neural network control unit 8 = NN
(= Second control circuit, neural network), FPS
O1 = Plant (= first floating body). And P
When only ID control is performed, the neural network
The control unit 8 does not operate, so that the sub FPSO control moment 11 is not output. The shuttle tanker is the second floating body.

The PID control unit 7 is a PID control device that outputs the control amount calculated by the PID control method based on the difference between the target value and the actually measured value. For example, in FIG.
, The target external force direction that is the direction of the external force (eg, the angle or direction between the tidal current (external force) and the base line of the coordinate system fixed to FPSO1, the angle between the tidal current (external force) and the X axis, or Azimuth, etc.) θ1 and actual FPSO1
Direction (feedback signal) θ2 Comparison result (θ1-
θ2) is input to the PID control unit 7. Then, the preset PID gain and the comparison result (θ1-θ
Based on 2) and, the main FPSO control moment 9 which is the controlled variable is calculated and output. External forces are complex natural environmental forces such as tidal, tidal, wind and wave forces.

The neural network control unit 8 includes a distance between the reference points of the FPSO 1 and the shuttle tanker 3 (eg, the distance between the end points DE of the connecting hawser 4 and the FPS).
The distance Ylen) R between the long axis determined by the bow and stern of O1 and the bow of the shuttle tanker 3 and the distance ΔR between the reference points are input every moment. In this embodiment, R
= The distance Y _len between the long axis determined by the bow and stern of FPSO1 and the bow of shuttle tanker 3 and ΔR = ΔY
It is _len . The neural network control unit 8 calculates the corrected turning angle Δθ (R, ΔR) obtained by the learning function based on the inputs of R and ΔR. Corrected turning angle Δ
θ (R, ΔR) is a turning angle conversion deviation between the FPSO 1 and the shuttle tanker 3 based on the relative distance and the relative distance deviation between the FPSO 1 and the shuttle tanker 3. The neural network control unit 8 uses the corrected turning angle Δθ (R,
The secondary FPSO control moment 11 is output based on (converting) ΔR). Further, the neural network control unit 8 has the minimum turning angle conversion value (turning angle conversion deviation) of the relative distance (Y _{len in} this embodiment) and relative distance deviation (ΔY _{len in} this embodiment) between the FPSO 1 and the shuttle tanker 3. Learn to be.

Main FPS output by PID control unit 7
The O control moment 9 and the sub-FPSO control moment 11 output by the neural network control unit 8 are
The target control moment 12 is additively input to the thruster group (not shown) of the FPSO 1. FPSO1 (Fig. 4
FPSO1 in the figure indicates that FPSO1 is controlled based on the target control moment 12 output by the control computer 6 and outputs the control result) is the relative angle (detection turning angle) θ2 of FPSO1 actually detected. Is output. θ2 is returned to the PID control unit 7 as the main feedback signal 13 and also to the neural network control unit 8 as the sub feedback signal 14. Neural network control unit 8
The feedback signal returned to is referred to as teacher data in neural network control. The output based on the operation of the ship higher engineer (captain, pilot) is the neural network control unit 8 as the teacher data 14 '.
Be fed back to.

Here, a learning method in the neural network control will be described. FIG. 5 shows the neural network control unit 8 in the control computer 6 of the FPSO 1 in FIG. The physical data of FPSO1 is used as the input data 21 for the external force (FPSO1).
Actuator, shuttle tanker actuator, hawser tension, natural environment force, total external force including external force acting on ship such as external force as reaction force of thruster force, or displacement (object fixed coordinate system (FPSO coordinate system)) lateral relative distance between the shuttle tanker at _{Y len,} variation [Delta] Y _len of the relative distance _{(Y le n),}
Etc.), output data (rotational motion and linear motion of surge, suway, yaw) 22 of a hull (including actuator (thruster)) that moves based on input data 21, and a neural net layer 23. . External force is
Generally, it is an external force that acts on each of the two buoyancy bodies, the FPSO1 and the shuttle tanker 3, but in this case, the shuttle tanker 3 is placed in a natural environment state only by being subjected to Hauser tension. There is.

In this embodiment, the input data 21 is
Displacement (lateral relative distance Y _len to the shuttle tanker in the object fixed coordinate system (FPSO coordinate system), relative distance (Y
_len ) change amount ΔY _len . Further, the output data 22 is the thrust of the thruster (actuator) of the FPSO 1, and is the sub-FPSO control moment 11 in FIG.
Is.

A plurality of neural net layers 23 are adopted when there are many control variables. The plurality of layers are a single or a plurality of input layers 23-1 and a single or a plurality of intermediate layers 23.
-2 and a single or a plurality of output layers 23-3. In this embodiment, the input layer 23-1 has the first output of the lateral relative distance Y _len and the change amount ΔY _len of the relative distance (Y _len ) with the shuttle tanker in the object fixed coordinate system (FPSO coordinate system). 24-1 and 24-2 are individually input to the first net node (node). First output 2
Each of 4-1 and 4-2 is multiplied by a coefficient, and is output from the first net node to the second output 25-1-1 to 25-1-2 and the second output 25.
-2-1 to 3 are output. Second output 25-1 to 2
To each of the three second net nodes of the intermediate layer 23-2. Each of the second outputs 25-1 and 25-2 is multiplied by a coefficient at the second net node, and output from the second net node as the third outputs 26-1 to 26-3. Each of the third outputs 26-1 to 26-3 has an output layer 23-.
Input to one third net node 3 of 3. Third output 26
Each of -1 to 3 is multiplied by a coefficient at the third net node, and the fourth output 27 is output from the third net node.
It is output as -1.

The input / output relationship of neurons (nodes) is expressed by the following equation.

[Equation 22] However, f (x) = tanh (x) has the following meanings. Q _i : output of the i-th neuron in a certain hierarchy ω _ij : coefficient from the i-th neuron to the j-th neuron in the next layer (weight value) u _j : j-th neuron in the next layer Input to neuron (sum of)

For learning the weight, a learning method based on the error back propagation method is used. When given teacher data, the weight value ω is set so as to minimize the evaluation function E based on the data.
_{If ij} is changed, the change amount Δω in the output layer
_ij is

[Equation 23] Becomes If E is the relative position,

[Equation 24] Since the proportional relation of is substantially established, it can be expressed by the following formula.

[Equation 25] The meaning of each code is as follows. η: learning coefficient

In FIG. 5, the fourth output 27-1 of the output layer 23-3 matches the teacher data 28 (or the evaluation function E (for example, the squared error between the output and the teacher data 28) is the minimum). A weight value ω _ij is calculated and determined in the network. If the number of neurons (the number of variables) is sufficiently large and the number of times teacher data is input to the output layer 23-3 is sufficiently large, the output data 22 corresponding to the input data 21 will asymptotically converge to a correct value. . The FPSO 1 that has completed such learning can acquire the sub feedback signal 14 that gradually approaches the target value according to the control signal that is the target control moment 12.

Neural network control unit 8
Has such a network structure. In this embodiment, the neural network control unit 8
The sub-feedback signal 14 fed back to
Secondary FPSO control moment 11 that itself outputs primary FPSO control moment 9 that PID control unit 7 outputs
It is the output value of FPSO1 corrected by. Then, the sub feedback signal 14 is referred to as teacher data.

Next, the operation of the floating body position control system of the present invention will be described. In the two-floating body system formed by the connecting hawser 4 of the FPSO 1 and the shuttle tanker 3 shown in FIG. 1, the system has an external force of sea and weather (tidal force, tidal force,
Complex natural environmental forces such as wind power and wave power are exerted. The FPSO1 measures a desired external force direction (ex. Tidal current (external force)) and a baseline of a coordinate system fixed to the FPSO1 from a device (not shown, which will be described later) for measuring those external forces that the FPSO1 has. Angle or azimuth between the tidal current (external force) and the X-axis, etc.) θ1,
Actual FPSO 1 direction (feedback signal, turning angle)
θ2 and are acquired.

Subsequently, in FIG. 4, the target external force direction θ1
Of the actual FPSO1 direction θ2 (θ1-θ
2) is input to the PID control unit 7. And
PID preset in the PID control unit 7
Based on the gain and the comparison result (θ1−θ2), the main FPSO control moment 9 that is the control amount is calculated and output.

When the control is only PID control, the main F
The PSO control moment 9 is output as it is to the control unit (included in FPSO1 in FIG. 4 of FIG. 4) that controls the movement of the actuator (pivot starter) as the FPSO swing thruster moment N _T that is the target control moment 12. It Then, in the above-described formula of the equation (8), N _T output from the PID control device and X _T ² + Y _T ² = T
T _S-F and δ _F are determined from the relationship of _S-F ² and the minimum control of the thruster thrust T _S-F . Then, by controlling T _S−F and δ _F of the swing thrusters of the FPSO 1, the moment control of the swing thrusters can be performed. The minimum control of thruster thrust T _S-F is
The minimum control of the thruster swing angle δ _F may be used.

In the equation (8), the PID control device
Output from N_TAnd X_T ^Two+ Y _T ^Two= T_SF ^Twoof
Solve the relationship and the equations of motion (1) to (3) in Equation 1.
T_SF, Δ_FIt is also possible to determine

After controlling the swing thruster, the target external force direction θ
The actual FPSO1 direction θ2 (turning angle), which is changed by control as 1 (because it is a natural force, may change from moment to moment), is measured again by the FPSO1. Then, PID control is continuously performed.

The control is PID control + neural network
In case of peak control, it becomes as follows. PID control unit
In P7, preset PID gain and comparison result
Based on (θ1-θ2), the main FPSO that is the control amount
The control moment 9 is calculated and output is the PID
This is similar to the case of only control. On the other hand, neural network
The work control unit 8 has a reference point distance R = FPSO.
Long axis and shuttle tanker determined by the bow and stern of 1
-Distance Y to the bow of 3 _lenAnd the momentary distance between reference points
Displacement amount ΔR = ΔY_lenAnd are entered. neural
The network control unit 8 is based on the inputs of R and ΔR.
And the corrected turning angle Δθ (R, ΔR) obtained by the learning function
To calculate. The corrected turning angle Δθ (R, ΔR) is FPSO
Reference point distance, which is the relative distance between 1 and shuttle tanker 3
Based on R and relative distance deviation ΔR which is relative distance deviation
Turning angle conversion between FPSO1 and shuttle tanker 3
Deviation. Neural network control unit 8
Converts the corrected turning angle Δθ (R, ΔR) into the sub FPS
The O control moment 11 is output.

The neural network control unit 8 converts the turning distance of the relative distance between the FPSO 1 and the shuttle tanker 3 (Y _{len in} this embodiment) and the relative distance deviation (ΔY _{len in} this embodiment) (turning angle conversion). deviation)
Is continuously learned during the control so that is minimized.
However, it is also possible to perform learning based on an appropriate learning pattern in advance and use the learning file that is the learning result for neural network control. The learning file includes data of ω _ij (coefficient (weight value) from the i-th neuron to the j-th neuron in the next layer).

Main FPS output by PID control unit 7
The O control moment 9 and the sub-FPSO control moment 11 output by the neural network control unit 8 are
The target control moment 12 is additively input to the thruster group (not shown) of the FPSO 1. θ2 is returned to the PID control unit 7 as the main feedback signal 13 and also to the neural network control unit 8 as the sub feedback signal 14. Since the subsequent process is the same as the PID control, the description thereof will be omitted.

Next, the PID control as described above or P
A simulation for carrying out a control method based on a combination of ID control and neural network control will be described. In this simulation, the description P
The position control of the two floating bodies of the FPSO 1 and the shuttle tanker 3 is simulated by a simulation device that simulates the same operation as the control computer 6 having the ID control unit 7 and the neural network control unit 8.

The control system of the control computer 6 shown in FIG. 4 is adjusted in terms of PID gain, learning coefficient, and addition coefficient. Referring to FIG. 7, a control parameter adjustment procedure of PID control and NN control in this embodiment will be shown. Note that this parameter adjustment procedure can be used not only in simulation but also in actual control. (1) Gain adjustment by PID control: Gain adjustment is performed by PID control. This is a basic control gain common to PID control and NN control. However, proportional (P) gain (K _P )
Are adjusted so that the FPSO tends to be directed toward the external force. "To show the tendency toward the direction of the external force" means that the control amount K based only on the proportional control is obtained from the difference (α1-α0) between the target value α0 and the measured value α1 in the controlled amount.
_This means that the magnitude of _P (α1-α0) is adjusted so as to indicate the direction of the subsequent change (for example, increase or decrease, plus or minus). Therefore, although it is made smaller than the control amount by other control, it is too small if it is negligible. In the present embodiment, for example, 0.1 (provided 0 ≦ K _P ≦ 1.0). The integral (I) gain (K _I ) is adjusted near zero. The response of the FPSO to the control is very slow, and whirling tends to occur when the integral control is increased, so the FPSO is made as small as possible. Considering that the gain is smaller than the proportional gain, the value is almost zero. For example, it is 0.01 or less. The derivative (D) gain (KD) is adjusted so that the FPSO does not swirl. The FPSO has a very slow response to control, but if the differential control is increased, whirling is less likely to occur, so the FPSO is increased. In this embodiment, for example, 1.0 (where 0 ≦ K _P ≦ 1.0). (2) Adjustment of learning coefficient of neural network: As a concrete method of learning of the neural network, it is proportional to the change amount ∂E / ∂ω _ij of the load, and this proportional constant is called a learning coefficient. The larger this coefficient is, the larger the amount of change per time is, and thus the learning speed becomes faster. However, if it is too large, the output becomes oscillating or saturated. Therefore, it is necessary to set the value within the range where such a thing does not occur. Also, as shown below, the inertia term (the second term on the right side)
Is effective in preventing vibration.

[Equation 26] The meaning of each code is as follows. η: learning coefficient α: inertia term coefficients η and α are set to almost the same value, and a value at which the output of the neural network does not vibrate or saturate is set. (3) Neural network addition coefficient adjustment: The output of the neural network 8 is -1 to +1 and needs to be converted because it is used as the moment control amount. Using the main FPSO control moment 9 output by the PID control unit 7 by PID control as a reference value, the conversion coefficient was adjusted by simulation so that the control result was improved. For example, if the ratio of the sub FPSO control moment 11 to the main FPSO control moment 9 is adjusted and the ratio is represented by k, the sub FPSO control moment = k (main FPSO control moment).
k is experimentally obtained.

Next, with reference to FIG. 24, the simulator 19 for performing the simulation will be described. The simulator 19 includes an input unit 15, a control calculation unit 16, a control computer unit 17, and an output unit 18. Input unit 15
Is an input device for inputting data related to the simulation. The control calculation unit 16 is a calculation device that performs a calculation related to the simulation based on the data input from the input unit 15 and a control algorithm stored in its own storage unit (not shown). In the present embodiment, calculations related to the motion of the floating body are performed. Inside the control computer unit 17
Based on the signal of the target control moment 12 from
An actuator simulation input circuit (not shown) for simulating the operation of the SO1 actuator (pivot thruster);
A floating body simulation output circuit (not shown) that outputs a motion value formed from the motion value of the PSO1 motion and the motion value of the shuttle tanker motion is provided. The calculation result is output to the control computer unit 17 or the output unit 19. The control computer unit 17 is an arithmetic unit that operates similarly to the control computer 6 in FIG. Based on the data input of the target external force direction θ1 and the actual FPSO1 direction θ2 from the control calculation unit 16, the calculation results by the PID control and the NN control are output to the control calculation unit 16 (actuator simulation input circuit (not shown)). To do. The output unit 18 is an output device for outputting a final simulation result.

Next, the operation (calculation procedure) of the floating body control simulator in the floating body control system of the present invention will be described with reference to FIGS. 6 and 24. In this embodiment, in the control calculation unit 16 and the control computer unit 17, the DPS
(Dynamic Positioning System
em) DP-MAP (Dyna) which is a dynamic analysis program.
mic Positioning System-Mo
simulation is performed using a relative position control program between two floating bodies, which is an extension of the Motion Analysis Program).

Here, the DOP-MAP will be described. This is a dynamic analysis program that uses an algorithm that determines the behavior of a floating body by establishing a motion equation of the floating body based on complex fluctuation forces such as wind, waves, and tidal currents, and solving it. Input the following input data, and consider the following output data obtained correspondingly. A1) Input data a1. Floating body summary data, b1. Environmental condition data, c1. External force coefficient data, d1. Mooring point data e1. Thruster data, f1. Control gain data g1. Simulation data B1) Output data h1. Time history, i1. Regarding the motion track input data, the floating body objective data is a value determined by the structure of the floating body. In this case, the target FPSO
The data of 1 is used. The environmental conditions and the external environmental force coefficient data are data relating to the external force of the sea condition and the weather that the target FPSO 1 receives now or in the future. The mooring point data is data relating to mooring point coordinates input when the FPSO 1 is moored. Thruster data is the thruster mounting position and maximum thruster capacity,
In this case, the FPSO1 thruster data is used. The control gain data is data used in a control arithmetic expression relating to PID control when PID control is performed on the movement of the FPSO 1. The simulation data is data required for other simulations. Regarding the output data, the time history is the external force of the FPSO1 due to the sea condition and the weather, the movement of the FPSO1 when the position holding control is performed under the external force (forward and backward, left and right, bow swing), the thruster swing angle in the position holding control, It is data representing changes in four items of thrust force in position holding control over time. In addition, the movement track is FPSO1 during the simulation period.
It is data showing a wake due to the movement of.

In the program for relative position control between two floating bodies based on the expanded DP-MAP in the present invention,
The above DP-MAP is a single floating body, but there are three major differences: handling two floating bodies, adding a Hauser model connecting the two floating bodies, and adding a control / learning routine by a neural network. . For, regarding the equation of motion, the equation of motion for two floating bodies is set up and calculated. With regard to, the tension of Hauser connecting the two floating bodies is introduced into the equation of motion. With regard to (1), the position control by the thruster is added to the PID control, and the control by the neural network is performed.

Accordingly, the main initial input data in the simulation are as follows. A2) a2. Floating body essential data (drainage, moment of inertia, mass, representative length, representative wind pressure area, draft, floating speed, hawser mounting position, hawser direction) b2. Environmental condition data (wave: wave direction / direction change, average wave period, significant wave height, regular wave / irregular spectrum, wind: wind direction change, wind speed, wind spectrum, tidal current: tidal current direction / direction change, tidal angle, tidal current Speed) c2. Environmental external force coefficient data (wave drift force coefficient, wind pressure coefficient, tidal current coefficient) d2. Mooring point data (mooring point coordinates) e2. Thruster data (thruster attachment point coordinates, maximum thruster capacity) f2. Control gain data (PID gain, calculation time interval, control time interval) g2. Simulation data

The output data in the simulation is
It becomes as follows. B2) Output data: h2. Time history- (external force, motion of floating body (including position): Surge, Sway, Yaw, swing thruster swing angle, swing thruster thrust (including rotation speed) Hauser tension) i2. Movement track

Now, a specific process of the operation of the floating body position control simulator (extended DOP-MAP) of the present invention will be described with reference to FIGS. 24 and 6. Step 1 (FIG. 6, S101) Input data read: Input data required for simulation (hereinafter, initial input data) via the input unit 15 in FIG. The initial input data is the above-mentioned data (a2-g2)
Enter. It is also possible to have the input unit 15 automatically read from an external storage device. Step 2 (S102) Program initialization: The program is initialized. Here, the neural network control unit (neural network control unit 8 in FIG. 4) in the control computer unit 17 is initialized.

Step 3 (S103) Computation of Fluid Force Acting on Two Floating Bodies: The control computing section 16 refers to the initial input data and computes the fluid force commonly acting on the two floating bodies. That is, FPSO1 and shuttle tanker 1
Is the fluid force commonly received by. It is a force related to the tidal current represented by each of the equations 2 to 5.

Step 4 (S104) FPSO equation of motion calculation: The control computing unit 16 refers to the initial input data and establishes the equation of motion for FPSO1 of the two floating bodies. That is, it is a motion equation related to the front-rear, left-right, and rotational motions of the FPSO 1 based on the forces by the thruster, wind, waves, and mooring lines. It is represented by each equation of the formula 1. Step 5 (S105) Fluid force acting on FPSO, mooring force calculation: In the control arithmetic unit 16, referring to initial input data, the thruster acting on FPSO1, wind, wave fluid force, and mooring line mooring force Is substituted into the above equation of motion and calculation is performed. The fluid force by the thruster is represented by the equation 8, the fluid force by the wind is represented by the equations 12 to 14, and the fluid force by the waves is represented by the equations 15 to 18. And
The following data is output to the control computer unit 17 out of the data on the movement of FPSO1 which is the result of the calculation. In the case of only PID control, the target external force direction (tidal current and FP
The angle or azimuth θ1 between the base line of the coordinate system fixed to SO1 and the actual external force direction (feedback signal) θ2 with respect to FPSO1 are output. In case of PID control + neural network control, θ1 and θ2
In addition, the distance between the reference points of the FPSO1 and the shuttle tanker 3 (the distance Y _len between the long axis determined by the bow and stern of the FPSO1 and the bow of the shuttle tanker 3) R, and the distance between the reference points at each moment ΔR (ΔR = ΔY _len ) is output. Step 6 (S106) Thruster control force calculation (neuro control amount calculation): PID preset by a PID control unit (not shown) inside the control computer unit 17
Based on the gain and the comparison result (θ1-θ2), the main FPSO control moment 9 that is the control amount is calculated. Further, the reference network distance R is controlled by a neural network control unit (not shown) inside the control computer unit 17.
(Y _len ), and the reference point distance displacement ΔR (Δ
Based on R = ΔY _len ), the sub FPSO control moment 11 is output. Then, the target control moment 12 obtained by adding the sub FPSO control moment 11 to the main FPSO control moment 9 is output to the control calculation unit 16. At this time, the neural network control unit uses the FPSO
1 and shuttle tanker 3 relative distance (Y in this embodiment)
_len ) and relative distance deviation (Δ in this embodiment)
Learning is performed so that the _turning angle conversion value (turning angle conversion deviation) of Y _len ) is minimized. Step 7 (S107) Convergence judgment: The control calculation unit 16
Simulation input circuit (not shown) inside
Then, the target control moment 12 is input to determine whether the position control of the FPSO 1 has succeeded. That is, the FPS
It is determined whether the O control moment is zero or less than a preset value. If yes, position control is successful,
That is, it is determined that the calculation has ended (= convergence). Then, the process proceeds to the next step (S108). If No, it is determined that the position control has not succeeded, that is, the calculation is in progress (= not converged).
Then, based on the FPSO control moment at that stage, the process from step 104 to step 106 is performed again. This is repeated until the calculation in step 106 converges. That is, the operation is repeated until the position control of the FPSO1 ends (the FPSO1 faces the target external force direction).

Step 8 (S108) Hauser tension calculation acting on the two floating bodies: After the position control of FPSO1 is completed,
The control calculator 16 calculates the hawser force exerted by the hawser tension on the FPSO 1 and the shuttle tanker 3 based on the equation (19).

Step 9 (S109) Shuttle tanker motion equation calculation: The control calculation unit 16 refers to the initial input data and the Hauser force to establish a motion equation for the shuttle tanker 3 of the two floating bodies. That is, it is a motion equation relating to the front-rear, left-right, and rotational movements of the shuttle tanker 3 based on the forces of propeller, thruster, rudder, wind, and waves. It is represented by each equation of the formula 1. Step 10 (S110) Calculation of fluid force acting on shuttle tanker: In the control computing unit 16, referring to the initial input data, the fluid force due to the propeller, thruster, rudder, wind, and wave acting on the shuttle tanker 3 is moved to the above motion. Substitute into the equation and perform the operation. The fluid force by the propeller is represented by the equation 7, the fluid force by the thruster is represented by the equation 9, the fluid force by the rudder is represented by the equation 11, the fluid force by the wind is represented by equations 12 to 14, and the fluid force by the wave is represented by equations 15 to 18. It Step 11 (S111) Convergence judgment: control calculation unit 16
Determines whether the shuttle tanker position calculation was successful. If Yes, that is, if the calculation is completed (= converged), the process proceeds to the next step (S112). If No, that is, if the calculation is still in progress (= not converged), the processes from step 109 to step 110 are performed again. This is repeated until the calculation in step 1110 converges.

Step 12 (S112) Convergence judgment: It is judged whether the simulation result has converged. If it has not converged, the process returns to S103 and the simulation is repeated. Step 13 (S113) Convergence judgment: It is judged whether the time set in the simulation has ended. If the time has not ended, the process returns to S103 and the simulation is repeated. END: With the end of step 13, the simulation ends.

The integration is performed using a numerical integration method such as the Runge-Kutta method.

Next, the result of the simulation test of the floating body position control simulation will be described with reference to the drawings. Here, the FPSO1 is assumed to be a modified model of a 120KDWAT tanker. In addition, shuttle tanker 3
We assumed a tanker of 68 KDEWT.

The external force condition set in the simulation is
It is as follows. Wave: H _1/3 = 4.5 m T _P = 6.5 sec. JONSWAP Spectrum β = 0 deg. Wind: None Tidal current: U _C = 2.0 m / s β _C = 15, −15 deg.

The simulation was performed under the following control conditions. No control: This is a free state where the thruster is not used and no control is performed. PID control: PID control is used to control the thruster. Neuro (always learning): A method of controlling thrusters by combining PID control and control by a neural network. Learning of the neural network is always performed simultaneously while controlling. Neuro (using learning file): Learning by a neural network is performed in advance in a zigzag pattern (see FIG. 8) to create a learning file. And P
A method of controlling thrusters by combining ID control and control by a neural network. A pre-learning file is used for control, and learning is not performed during control.

The results of the simulation will be described with reference to FIGS. Each figure shows the time history of the hawser tension and the distance between the two floating bodies under the control conditions 1 to 3 above. In each figure, the horizontal axis is the simulation time (seconds), the vertical axis is the upper graph, and Hauser tension (to
n), the lower graph is the distance between two floating bodies (m).

In the case without the control shown in FIG. 9, the hawser tension of 150 to
The number of times is n or more and the time is large. In addition, the number of times and the time when the distance between the two floating bodies becomes -20 m or less is large.

In the case of the PID control of FIG. 10, as compared with the case without control, both the number of times and the time when the hawser tension is 150 tons or more and the number and the time when the distance between the two floating bodies is -20 m or less, Compared to the case without control,
is decreasing. That is, it was revealed that the PID control has the effect of reducing the hawser tension and the distance between the two floating bodies and stabilizing the same.

When the neural network control for always learning is added to the PID control of FIG.
Compared to the case of control, the hawser tension will not exceed 150 tons, and the maximum hawser tension will be about 130 tons.
It has dropped to n. Furthermore, the effect of constant learning is that the hawser tension is 100 tons when the learning progresses for 1500 seconds or more.
It can be controlled below. If you extend the time further,
It is considered possible to set it to 50 tons or less. Also,
The number of times and the time when the distance between the two floating bodies becomes -20 m or less are reduced, and the variation is stable in the range of approximately -5 m to -25 m. That is, as compared with the case of the PID control, it was confirmed that the effect of reducing the hawser tension and the distance between the two floating bodies by adding the neural network control for always learning and the stabilization thereof was added.

When the pre-learned neural network control is added to the PID control shown in FIG.
Compared to the case of D control, the hawser tension never exceeds 150 tons, and the maximum hawser tension is about 130 t.
It has decreased to on and 100 ton or less has increased as a whole. Furthermore, the distance between the two floating bodies is always very stable in the range of -10 m to -20 m except the initial stage. That is, as compared with the case of the PID control, the effect of reducing the hawser tension and the distance between the two floating bodies by adding neural network control for performing the pre-learning and its stability can be confirmed.

From the above results, it is possible to favorably control the hawser tension and the distance between the two floating bodies by introducing the PID control. Further, by adding neural network control to the PID control, it becomes possible to better control the hawser tension and the distance between the two floating bodies. In particular, constant learning of neural network control makes it possible to control the hawser tension to a very small value. In addition, in particular, it is possible to perform the control for keeping the distance between the two floating bodies stably within a certain range by preliminarily learning the neural network control.

In the neural network control, in order to perform a wide range of generalization, a method combining the above-mentioned control methods and is also an effective control method. In other words, it is a control system in which neural network control is added to PID control, and a method of performing learning in advance on the neural network and continuously performing learning even during control.

Next, as a simulation of two floating bodies,
An aquarium test was conducted to analyze the characteristics of the FPSO 1 and the shuttle tanker 3. The FPSO1 can perform both PID control and neural network control.
The characteristic data of the FPSO 1 and the shuttle tanker 3 are collected by the water tank test for the characteristic analysis of the two-floating body problem.
The identification of the characteristic data is performed by a neural network unique to FPSO1. The neural network learns so that the dynamic characteristics of the aquarium test match the output 22 of the neural network.

FIG. 14 shows a test device for the water tank test. Overall, the tests were carried out on a 1/50 scale size of the real thing. The test device for the water tank test includes a measurement carriage 31, a towing jig 44, an FPSO compatible model 32, a hawser compatible simulated spring 34, and a ship position measuring device 45. The measurement carriage 31 includes a PC 41 for FPSO control and a P for shuttle control.
It is equipped with C42 and PC43 for data analysis. FP
The SO-compatible model 32 is a swinging thruster 38 for turning angle control.
It is equipped with. The shuttle-compatible model 33 includes a tag-pulling force-compatible simulated blower 35, a rudder 36, a propeller 37, and a bow thruster 39.

The measurement carriage 31 is an FP that is a two-body floating structure for testing.
It is a device that captures various data while controlling and towing the SO compatible model 32 and the shuttle compatible model 33. FPS
The O-controlling PC 41 is a personal computer (PC) that controls the operation of the FPSO-compatible model 32 and takes in data regarding the FPSO-compatible model 32. F
PSO model azimuth, FPSO model azimuth velocity,
Handles information related to control such as the FPSO compatible model side hawser tension, FPSO compatible model thruster swing angle, and relative position with the shuttle compatible model. PC42 for shuttle control
Controls the operation of the shuttle compatible model 33,
This is a PC that captures data regarding the shuttle-compatible model 33. Handles control-related information such as shuttle compatible model O azimuth, shuttle compatible model azimuth angular velocity, shuttle compatible model side hawser tension, FPSO thruster swing angle, and relative position with the FPSO compatible model. PC43 for data analysis
PC41 for FPSO control and PC42 for shuttle control
This is a PC that analyzes the motion of the two-float body based on the data relating to the two-float body captured by and.

One end of the towing jig 44 is the measurement carriage 31.
The other end is connected to the FPSO compatible model 32.
The measurement carriage 31 then tows the FPSO-compatible model 32 in a fixed direction via the towing jig 43. The FPSO compatible model 32 is a model ship (about 5 m) simulating FPSO (about 250 m). In order to control its own turning angle, it has a turning angle control swinging thruster 38 at its stern. The shuttle-compatible model 33 is a shuttle tanker (about 2
It is a model ship (about 5 m) simulating 50 m). Bow thruster 39 (next to ship side) used for navigation and position control
On the bow, and a propeller 37 and its rudder 36 on the stern. Further, a tag-pulling force-compatible simulated blower 35 is mounted on the stern side of the shuttle-compatible model 33. The hawser-compatible simulated spring 34 is a spring simulating a hawser that connects the stern of the FPSO-compatible model 32 and the bow of the shuttle-compatible model 33. The stern of the FPSO compatible model 32 and the bow of the shuttle compatible model 33 are set as both end points of the relative distance described above. Both models are connected to the Hausa corresponding simulated spring 34, floated on the water surface (sea water surface) of the water tank, and are towed by the measurement cart 31, and no ocean current (tidal current) is generated in the water tank. The ship position measuring device 45 is a position measuring device capable of accurately measuring the positions of the FPSO compatible model 32 and the shuttle compatible model 33.

Through the water tank test, the FPSO compatible model 32
The shuttle-compatible model 33 is trained so that the output as a motion with respect to an external force matches the teacher data.

FIG. 15 and FIG. 16 are tables showing a list of test conditions for the control water tank test shown in FIG. There are four control modes. The four types are PID control (PID control).
Control is performed: FIG. 15) and PID control + neural network control (always learning) (PID control and neural network control (always learning) are performed simultaneously: FIG. 16).
And PID control + neural network control (learning file) (PID control and neural network control (preliminary learning) are performed simultaneously: FIG. 16). The test conditions are shown for each control mode.
The controlled object is the FPSO azimuth. The test conditions are wave height, tidal current velocity, tidal current direction, tag pulling force, and F
PSO is thruster swing angle and thruster maximum thrust, and shuttle 33 is bow thruster maximum thrust and propeller thrust. In this case, the shuttle 33 is discharged in a free state (maximum thrust of bow thruster = propeller thrust = 0). The leftmost columns of the tables in FIGS. 15 and 16 are
The test number is shown.

In the case of neural network control
FPSO stern-shuttle
The relative position of the bow in the Y direction and its variation (Y_len,
ΔY _len) Is used.

FIG. 13 shows the coordinate system in the water tank test.
FIG. 13 shows the FPSO compatible model 32 and the shuttle compatible model 3.
3 shows the definition of the measurement target amount of the physical / geometrical data of 3 (output data as the data of the FPSO-compatible model 32 and the shuttle-compatible model 33 itself). Figure 13
The quantities to be measured are shown for each of the FPSO compatible model 32 and the shuttle compatible model 33. The meaning of each code is as follows. The description in parentheses after the description of each measurement target amount indicates a measuring device suitable for measuring the measurement target amount. They are FPSO compatible models 32
And a shuttle-compatible model 33 or a ship position measuring device 45 (not shown) to measure and measure the FPSO control PC.
Data is output to 41 and the shuttle control PC 42. δ _F : FPSO thruster swing angle (potentiometer), n _S-F : FPSO thruster propeller rotation speed (tachometer), ψ _F : FPSO azimuth angle (gyro), r _F : FPSO azimuth angular velocity (gyro), φ _F : FPSO roll angle (tilt meter), θ _F : FPSO pitch angle (tilt meter), X _0-T : FPS OX direction mooring position (potentiometer), Y _0-T : FPS YO direction mooring position (potentiometer), X _{0- F} : FPSO stern target X direction position (optical tracker), Y _0-F : FPSO stern target Y direction position (optical tracker), F _X : FPSO turret X direction mooring force (block gauge), F _Y : FPSO turret Y direction mooring force (block _gauge), T S-F: FPSO thruster thrust (block gauge), T : FPSO-S / T Hosa tension (tensiometer), ψ _H-F: FPSO side Hosa tension relative angle (potentiometer), _{H W:} height (height gauge), δ _{S: S} / T steering angle (potentiometer), n _P-S: S / T propeller speed _{(tachometer), n S-B: S} / T bow thruster rotational speed (tachometer), ψ _{S: S} / T azimuth _(gyro), r S: S / T orientation Angular velocity (gyro), φ _S : S / T roll angle (tilt meter), θ _S : S / T pitch angle (tilt meter), X _0-S : S / T stern target X direction position (optical tracker) ), Y _0-S : S / T stern target position in Y direction (optical tracker), F _N : S / T rudder direct pressure (block gauge), _FT : S / T rudder parallel force (block gauge), T _P: S / T propeller thrust (load _{cell), T S-B: S} / T bow thruster thrust (block gauge), ψ _TUG : S / T tag tension direction (potentiometer) ψ _H-S : S / T side hawser tension relative angle (potentiometer) (S / T: shuttle tanker)

In the test, waves and tidal current were considered as external forces. The waves are generated by the wave generator on the short side of the water tank. For the tidal current, the tidal current velocity and direction were simulated by operating the measurement carriage 31 at the set velocity and moving the two floating bodies. That is, the tidal current is replaced by the relative linear movement of the measurement carriage 31.

With respect to the control method, the shuttle tanker side conducted the test without control, so the actuators (propeller 37 and bow thruster 39) of the shuttle tanker were not operated. On the FPSO side, the personal computer on the measurement carriage 31 feeds back the output from each sensor to perform data processing to control the thruster swing angle and the thruster rotation speed of the swing thruster 38. In addition, with a blower simulating the tag pulling force, the shuttle tanker azimuth angle was fed back so that a constant pulling force was always applied in the space fixed coordinate X-axis direction, and the tag pulling force direction control was performed.

The external force conditions set in the test are as follows. Wave: Significant wave height H _1/3 = 4.5 m, wave height peak period T _P = 6.5 sec. Corresponding to 3.5 m. Spectrum = JONSWAP Spectrum direction distribution function = cos ⁴ β wave direction φ _W = 0 deg. Wind: None Tidal current (simulation): Velocity U _C = 2kn, velocity corresponding to 3kn Tidal current direction φ _C = 25, 15, 5, -15 deg. (“Corresponding to height (speed)” means that the height (speed) is obtained in an actual FPSO or shuttle tanker.) The azimuth target is PID control: φset = 0 deg. , Neural network: φset = 0 deg. Is.

17 to 20 numerically show the relative movements of the FPSO corresponding model 32 and the shuttle corresponding model 33 for the four test numbers. The horizontal axis is the time history (seconds) of the simulation. The vertical axis is as follows in order from the top of the figure in which 10 pieces are arranged. The description in parentheses after the description of the vertical axis indicates the correspondence between these and the symbols in FIG. 13. DELTA-F (deg): FPSO thruster swing angle (δ _F ), NS-F (rpm): FPSO thruster propeller rotation speed (n _S-F ), PSI-F (deg): FPSO azimuth angle (ψ _F ). , TS-F (ton): FPSO thrusters thrust _{(T S-F) HWTEN:} FPSO-S / T Hosa tension _{(T H), X0-F} : FPSO stern target X-direction position (X
_0-F ), Y0-F: FPSO stern target position in Y direction (Y
_0-F ), PSI-S (deg): S / T azimuth (ψ _S ), X0-S: S / T stern target X direction position (X _0-S ), Y0-S: S / T stern target Y direction position (Y _0-S ), (S / T: shuttle tanker)

FIG. 17 corresponds to test number 8, FIG. 18 corresponds to test number 13, FIG. 19 corresponds to test number 22,
FIG. 20 corresponds to test number 30. 15 and 1
As shown in 6, test number 8, test number 13, test number 22, and test number 30 are the same in terms of test conditions. However, they are different in the test mode, the test number 8 is no control, the test number 13 is PID control, the test number 22 is PID control + neural network control (always learning), and the test number 30. Is PID control + neural network control (learning file).

In the non-control mode of FIG. 17, the swing thruster of the FPSO is inoperative, but in the PID control mode of FIG. 18, the swing thruster of the FPSO is violently operating. In the second half of the time history, where the transient response is considered to be calm and to show valid data for the experiment, the lateral displacement Y0-F of the FPSO has reduced fluctuations in the PID control mode and is stable compared to the non-control mode. I understand that. However, within this time history range, there is no significant difference in hawser tension (HWTEN). This is because the disturbance is large (equivalent to a wave height of 4.5 m). That is, although the PID control is effective, when the disturbance is large (equivalent to a wave height of 4.5 m), the effect is small.

In the PID control + neural network control (always learning) mode of FIG. 19, the second half of the time history, which is considered to show effective data as an experiment, with a stable transient response under the same disturbance as in FIGS. 17 and 18. It can be seen that in the portion, the fluctuation of the lateral displacement Y0-F of the FPSO is reduced and is stable as compared with the non-control mode. In addition, both the average value and the peak value of the hawser tension (HWTEN) are reduced as compared with the no control mode and the PID control mode. That is, the effectiveness of adding neural network control (always learning) is shown.

In the PID control + neural network control (learning file) mode of FIG. 20, FIG. 17 and FIG.
In the latter half of the time history, where the transient response settles down under the same disturbance as 8 and shows valid data as an experiment, the fluctuation of the lateral displacement Y0-F of the FPSO decreases, and it is compared with the uncontrolled mode. You can see that it is stable. In addition, both the average value and the peak value of the hawser tension (HWTEN) are reduced as compared with the no control mode and the PID control mode. That is, the effectiveness of adding the neural network control (learning file) is shown. However, the control performance is slightly inferior to the PID control + neural network control (always learning) mode.

21 to 23 show changes in hawser tension depending on the control method in relation to the wave height. The hawser tension is the average and highest value in the time history. The horizontal axis indicates each control method, and the vertical axis indicates Hauser tension (HW
TEN).

FIG. 21 (a) shows the Hauser tension in each control method at a wave height of 3.5 m, and FIG. 21 (b) shows
It is the Hauser tension by each control method at a wave height of 4.5 m.

In FIG. 21 (a), when the wave height is equivalent to 3.5 m, the maximum value is suppressed even in the PID control mode, and it is possible to suppress fluctuations in the hawser tension.
That is, it is more effective in suppressing the hawser tension than in the control-free mode. Further, when the neural network control is added thereto, the maximum value is about the PID control, but the average value is lowered. It is possible to further suppress fluctuations in hawser tension. That is, it is more effective in suppressing the hawser tension than in the PID control mode.

In FIG. 21B, when the wave height is equivalent to 3.5 m, the PID control mode is almost the same as the non-control mode. That is, when the disturbance is large, the PID control alone is not effective in suppressing the hawser tension. However, adding neural network control to it lowers the maximum and average values. It is possible to further suppress fluctuations in hawser tension. That is, P
This is more effective in suppressing the hawser tension than in the ID control mode.

FIG. 22A shows a wave height of 4.5 m and a tide direction of 25 deg. It is the Hauser tension in each control method in. 22B shows comparison data with FIG. 22A and is the same as FIG. 21B. Wave height equivalent to 4.5 m and tidal current direction 15 deg. It is the Hauser tension in each control method in.

In FIG. 22 (a), when compared without control, fluctuations in hawser tension can be kept low and maximum and average values can be reduced in both PID control mode and PID control + neural network control. That is, it is more effective in suppressing the hawser tension than in the control-free mode.
However, there is no significant difference between the PID control mode and the PID control + neural network control. It is considered that this is because the neural network cannot sufficiently learn due to the constraints of the test equipment.

FIG. 23 (a) shows hawser tension in each control method when the wave height is 4.5 m and the tag pulling force is 15 tons. FIG. 23B shows comparison data with FIG. 23A and is the same as FIG. 21B. It is the Hauser tension in each control method when the wave height is equivalent to 4.5 m and the tag pulling force is 30 tons.

In FIG. 23 (a), comparing without control, good Hauser tension control is performed in the order of PID control + neural network control (learning file), PID control + neural network control (always learning), and PID control mode. It is possible to do. That is, it is more effective in suppressing the hawser tension than in the control-free mode.

The following effectiveness was confirmed by utilizing PID control and neural network control for the movement of the two-floating body composed of the shuttle tanker connected by FPSO and hawser. That is, compared to the case of no control, PI
The D control is effective for controlling the two floating bodies, and can suppress the swing of the FPSO and reduce the hawser tension. When neural network control is added to the PID control, it is more effective for controlling the two floating bodies, and further, the swing of the FPSO can be suppressed and the hawser tension can be reduced.

The basic motion of the ship is divided into three motions of surge, sway and yaw. A nonlinear equation of motion holds for each of the three movements. A three-dimensional external force is given to each as an external force. From the three nonlinear equations of motion, the acceleration can be calculated for the three surge, sway, and yaw motions. The accelerations ua, va, ra with respect to the three-dimensional coordinate axes u, v, r are generally represented by the following equations. ua = f (v, r, u, Xj) va = g (v, r, u, Yj) ra = h (v, r, u, Nj) where Xj, Yj, Nj are surge, sway, These are multiple external forces related to yaw. If the accelerations ua, va, ra are given as teacher data, the non-linear functions f, g, h can be obtained by the neural network.

[0143]

The relative floating body motion system according to the present invention,
In the motion control circuit and simulator between two floating bodies, the feedback control amount of the motion control between the two floating bodies that mechanically affects each other is effectively corrected by the control amount of the neural network, and it is possible to control the nonlinear motion which was very difficult in the past. It can be effective and easy.

[Brief description of drawings]

FIG. 1 is a perspective view showing an application example of an embodiment of a relative floating body movement system according to the present invention.

FIG. 2 is a diagram showing a coordinate system of a mathematical model related to the relative floating body motion system according to the present invention.

FIG. 3A is a diagram showing an example of a circuit block diagram relating to an embodiment of a motion control circuit between two floating bodies according to the present invention. (B) It is a figure which shows an example of the circuit block diagram of embodiment of the movement control circuit between two floating bodies by this invention. (C) It is a figure which shows an example of the circuit block diagram of embodiment of the motion control circuit between two floating bodies by this invention.

FIG. 4 is a circuit block diagram showing an embodiment of a relative floating body movement system according to the present invention.

FIG. 5 is a diagram showing a configuration of a neural network of an embodiment of a motion control circuit between two floating bodies according to the present invention.

FIG. 6 is a diagram showing a flow of a floating body position control program according to the present invention.

FIG. 7 is a diagram showing a procedure for adjusting control parameters of a floating body position control program according to the present invention.

FIG. 8 is a diagram showing a pre-learning pattern of the neural network of the floating body position control program according to the present invention.

FIG. 9 (a) is a diagram showing hawser tension in a simulation result (in the case of no control) of the floating body position control program according to the present invention. (B) It is a figure which shows the distance between two floating bodies among the simulation results (in the case of no control) of the floating body position control program by this invention.

FIG. 10 (a) is a diagram showing hawser tension in the simulation result (in the case of PID control) of the floating body position control program according to the present invention. (B) It is a figure which shows the distance between two floating bodies among the simulation results (in the case of PID control) of the floating body position control program by this invention.

FIG. 11 (a) is a diagram showing hawser tension in a simulation result (in the case of PID control + neural network control (always learning)) of the floating body position control program according to the present invention. (B) It is a figure which shows the distance between two floating bodies among the simulation results (in the case of PID control + neural network control (constant learning)) of the floating body position control program by this invention.

FIG. 12 (a) is a diagram showing hawser tension in the simulation result (in the case of PID control + neural network control (learning file)) of the floating body position control program according to the present invention. (B) It is a figure which shows the distance between two floating bodies among the simulation results (in the case of PID control + neural network control (learning file)) of the floating body position control program by this invention.

FIG. 13 is a plan view showing a test measurement amount in a water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 14 is a front view with a plan view showing the configuration in the water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 15 is a table showing the test conditions of the water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 16 is a table showing test conditions of a water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 17 is a graph showing test results of a water tank test relating to the embodiment of the floating body position control system according to the present invention.

FIG. 18 is a graph showing another test result of the water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 19 is a graph showing still another test result of the water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 20 is a graph showing another test result of the water tank test according to the embodiment of the floating body position control system according to the present invention.

FIG. 21 (a) is a graph showing the analysis result of the water tank test according to the embodiment of the floating body position control system according to the present invention. (B) It is a graph which shows the comparative data of the analysis result of the water tank test regarding embodiment of the floating body position control system by this invention.

FIG. 22 (a) is a graph showing another analysis result of the water tank test according to the embodiment of the floating body position control system according to the present invention. (B) It is a graph which shows the comparative data of the other analysis result of the water tank test regarding embodiment of the floating body position control system by this invention.

FIG. 23 (a) is a graph showing still another analysis result of the water tank test according to the embodiment of the floating body position control system according to the present invention. (B) It is a graph which shows the comparative data of the further another analysis result of the water tank test regarding embodiment of the floating body position control system by this invention.

FIG. 24 is a diagram showing a configuration of an embodiment of a floating body position control simulator according to the present invention.

FIG. 25 is a diagram showing a configuration of a hydraulic device in the embodiment of the floating body position control system of the present invention.

[Explanation of symbols]

1 FPSO 1'floating body 2 mooring lines 3 shuttle tankers 4 Hausa 6 control computer 7 PID control unit 7'controller 8 Neural network control unit 8'Neural network control unit 9 Main FPSO control moment 11 Sub FPSO control moment 12 Target control moment 13 Main feedback signal 14 (14 ') Sub feedback signal 15 Input section 16 Control calculation unit 17 Control computer section 18 Output section 19 Simulator section 21 Input data 22 Output data 23 Neural Net Layer 23-1 Input layer 23-1 Middle layer 23-1 Output layer 24 1st output 25 Second output 26 3rd output 27 Fourth output 31 Measuring cart 32 FPSO compatible model 33 Shuttle compatible model 34 Hausa compatible simulated spring 35 Tag Towing force compatible simulated blower 36 rudder 37 Propeller 38 Swing thruster for turning angle control 39 Propeller 41 FPSO control PC 42 Shuttle control PC 43 Data analysis PC 44 Towing jig 45 Ship Position Measuring Device 100 connections 100-1 Connection 100-2 connection 101 Hydraulic control unit 102 hydraulic system

Front page continuation (51) Int.Cl. ⁷ Identification code FI theme code (reference) G05B 13/02 G05B 13/02 L G06F 19/00 110 G06F 19/00 110 (72) Inventor Masami Matsuura Nagasaki City, Nagasaki Prefecture 5-717-1, Fukahori-machi Sanryo Heavy Industries Co., Ltd. Nagasaki Research Institute (72) Inventor Eiichi Kobayashi 5-717-1, Fukahori-machi Nagasaki City, Nagasaki Sanryo Heavy Industry Co., Ltd. F-term (reference) 5H004 GA15 GB14 HA07 HA10 HA16 HB07 HB08 JA22 JB08 JB18 JB30 KA65 KA71 KB02 KB04 KB06 KD36 KD45 KD47 LA15 5H303 AA11 BB02 BB09 BB14 BB20 EE03 JJ05 KK02 KK03 KK04 KK11 KK21 MM05 QQ08 Q09 Q09

Claims (17)

[Claims] 1. A first floating body, and a second floating body displaceably connected to the first floating body, wherein the first floating body acts on the first floating body in a direction of the first floating body. A floating body position control system, comprising: a PID control circuit that controls the movement of the first floating body so as to direct the first floating body to the direction of the external force based on the direction of the external force. 2. The gain of the PID control circuit, the proportional gain is adjusted so as to show a tendency to move the first floating body in the direction of the external force, the integral gain is adjusted to near zero, and the differential gain is The floating body position control system according to claim 1, wherein the first floating body is adjusted so as not to cause whirling. 3. A first floating body, and a second floating body displaceably connected to the first floating body, wherein the first floating body comprises a control circuit for controlling the movement of the first floating body. A first control circuit that outputs a first output signal based on an input of a first input signal that is a physical quantity related to position control of the first floating body; and the first floating body with respect to the second floating body. A second control circuit that outputs a second output signal based on an input of a second input signal that is a relative physical quantity of the first floating body, and the movement of the first floating body includes the first output signal and the second output. A floating body position control system, wherein the second control circuit is a neural network that outputs the second output signal for correcting the first output signal. 4. The first signal is a direction of an external force acting on the first floating body with respect to the first floating body, and the second signal is a relative distance between the first floating body and the second floating body. The floating body position control system according to claim 3, wherein the floating body position control system is an amount based on the displacement and the amount based on the displacement of the relative distance. 5. A first floating body, and a second floating body displaceably connected to the first floating body, wherein the first floating body is a physical quantity relating to position control of the first floating body. A floating body position control system, comprising: a first control circuit that controls the movement of the first floating body based on one input signal, wherein the first control circuit is a neural network. 6. A first floating body, and a second floating body displaceably connected to the first floating body, wherein the first floating body comprises a control circuit for controlling the movement of the first floating body. A first control circuit that outputs a first output signal based on an input of a first input signal that is a physical quantity related to position control of the first floating body; and the first output signal and the first floating body. A second control circuit that outputs a second output signal based on an input of a second input signal that is a motion output of the motion of the first floating body is controlled based on the first output signal, The floating body position control system, wherein the second control circuit is a neural network that outputs the second output signal for correcting the output signal of the first control circuit. 7. The floating body position control system according to claim 5, wherein the first signal is a direction of an external force acting on the first floating body. 8. The floating body according to claim 3, wherein the motion output of the first floating body is returned to the neural network as teacher data, and the neural network normally performs learning based on the teacher data. Position control system. 9. The floating body position control system according to claim 3, wherein the neural network controls the first floating body after performing learning in advance. 10. The floating body position control system according to claim 3, wherein the first control circuit is a PID control circuit. 11. The floating body position control system according to claim 3, wherein the second floating body is connected to the first floating body via a hawser. 12. The neural network comprises multiple layers, wherein the i-th neuron output of one layer of the multiple layers is represented by Q _i , and is output to the j-th neuron of the next layer of the one layer. An input neuron is represented by Q _j , a weighting factor between the i-th neuron and the j-th neuron is represented by ω _ij , and the neural network has the following equation: Q _j = f (Σ The relative floating body motion system of claim 3 represented by _i ω _ij Q _i ). 13. The relative floating body motion system according to claim 12, wherein the expression is represented by Q _j = tanh (Σ _i ω _ij Q _i ). 14. An actuator simulation input circuit to which an operation signal for operating an actuator of a first floating body is input, a first motion value of a motion of the first floating body, and a first displacement value movably connected to the first floating body. 2 a floating body simulation output circuit that outputs a motion value formed from a second motion value of the motion of the floating body; and an actuator simulation input circuit and the floating body simulation output circuit that receive the motion signal and output the motion value. A floating body position control simulator including a neural network interposed therebetween. 15. The neural network comprises multiple layers, wherein the i-th neuron output of one layer of the multilayer is represented by Q _i , and is output to the j-th neuron of the next layer of the one layer. An input neuron input is represented by Q _j , a weighting factor between the i-th neuron and the j-th neuron is represented by ω _ij , and the neural network has the following equation: Q _j = tanh (Σ _15. The floating body position control simulator according to claim 14, represented by _i ω _ij Q _i ). 16. A step of generating a first equation of motion which is a equation of motion of a first floating body, a step of calculating a first external force which is a total of forces exerted on the first floating body, and the first equation of motion. Solving the first equation of motion from the first external force; generating a second equation of motion that is the equation of motion of a second floating body that is displaceably connected to the first floating body; Calculating a second external force, which is the total force exerted on the floating body, solving the second equation of motion from the second equation of motion and the second external force, the first floating body and the second floating body And a step of calculating the tension of the connected hawser that connects the and. 17. A program for executing the floating body position control simulation method according to claim 16.