Dynamic positioning and thrust distribution device and method based on artificial neural network

Patent metadata

Field	Value
Publication number	US-11500339-B2
Application number	US-201916764665-A
Country	US
Kind code	B2
Filing date	Dec 23, 2019
Priority date	Dec 26, 2018
Publication date	Nov 15, 2022
Grant date	Nov 15, 2022

Abstract

The invention provides a dynamic positioning and thrust distribution method based on an artificial neural network, is a quadratic programming problem in the optimization problem, and can compute the thrust coefficient of a rear thruster in a constraint condition of a thrust distribution problem by taking into account of the corner of a front thruster and through the artificial neural network. Then the optimization problem enabling the power of the thruster to be minimized is solved according to a sequential quadratic programming algorithm, so that a thrust distribution scheme on the azimuth thrusters is obtained. Meanwhile the invention further provides a dynamic positioning and thrust distribution device based on an artificial neural network. The invention, through the introduction of the concept of thrust coefficients, on the one hand, can accurately quantize thrust loss, and on the other hand can enlarge the feasible area of the rotation angle of the thruster, thereby ensuring that the more optimized and reasonable result can be obtained for the quadratic programming problem, reducing the power of the thruster and saving energy.

First claim

What is claimed is:

1. A dynamic positioning and thrust distribution method based on an artificial neural network, characterized by comprising the following steps:
Step S1: Establish and train a fitting thrust coefficient of an artificial neural network in a real-time control computer;
Step S2: Add the trained artificial neural network into a thrust distribution model to obtain the following model:

min
⁢

∑

i
=
1

8

⁢

c
i

⁡

(

ρ
,
D

)

·

T
i

3
2

∑

i
=
1

8

⁢

T
i

·
cos

⁢

⁢

α
i

·

η
i

-

F
x

=
0

∑

i
=
1

8

⁢

T
i

·
sin

⁢

⁢

α
i

·

η
i

-

F
y

=
0

∑

i
=
1

8

⁢

[

T
i

·

η
i

⁡

(

x
i

·
sin

⁢

⁢

α
i

-

y
i

·
cos

⁢

⁢

α
i

)

]

-

M
z

=
0

T
min

≤

T
i

≤

T
max

Wherein, c i , is a constant of each azimuth thruster, which is related to fluid density p and propeller diameter D, T max , T min respectively indicate the upper limit and lower limit of the thrust of each azimuth thruster, and F x ,F y ,M z are separately resultant force and resultant moment of all the thrusters in three freedom degree directions which are surging, swaying and yawing directions;
Step S3: Set the resultant force and resultant moment of all the thrusters in the three freedom degree directions which are surging, swaying and yawing directions;
Step S4: Perform thrust distribution iteration algorithm according to a quadratic programming algorithm, Thrust distribution mathematical model is briefly recorded as:

{

min
⁢

⁢

f
⁡

(
x
)

Assignees

• Univ Shanghai Jiaotong

Inventors

• Wang Lei
• Li Bo
• Yi Fan
• Wang Xuefeng
• He Huacheng
• Qin Licheng

Classifications

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title

• G05B13/027 Primary

  using neural networks only · CPC title

• G05B13/042 Primary

  in which a parameter or coefficient is automatically adjusted to optimise the performance · CPC title

• G06N3/04

  Architecture, e.g. interconnection topology · CPC title

• G06N3/0499

  Feedforward networks · CPC title