CN112084574B - Ship additional mass and additional moment of inertia determining method based on neural network - Google Patents
Ship additional mass and additional moment of inertia determining method based on neural network Download PDFInfo
- Publication number
- CN112084574B CN112084574B CN202010800265.5A CN202010800265A CN112084574B CN 112084574 B CN112084574 B CN 112084574B CN 202010800265 A CN202010800265 A CN 202010800265A CN 112084574 B CN112084574 B CN 112084574B
- Authority
- CN
- China
- Prior art keywords
- ship
- additional
- additional mass
- inertia
- mass
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 55
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 21
- 238000012360 testing method Methods 0.000 claims abstract description 51
- 239000011159 matrix material Substances 0.000 claims abstract description 22
- 238000004088 simulation Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000002474 experimental method Methods 0.000 claims abstract description 14
- 230000010355 oscillation Effects 0.000 claims description 51
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000012937 correction Methods 0.000 claims description 8
- 238000010998 test method Methods 0.000 claims description 7
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 7
- 239000003255 cyclooxygenase 2 inhibitor Substances 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 11
- 230000006870 function Effects 0.000 description 13
- 238000013016 damping Methods 0.000 description 8
- 238000006073 displacement reaction Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000013139 quantization Methods 0.000 description 5
- 230000001133 acceleration Effects 0.000 description 4
- 238000013459 approach Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000012530 fluid Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000010561 standard procedure Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Computer Hardware Design (AREA)
- Mathematical Analysis (AREA)
- Aviation & Aerospace Engineering (AREA)
- Automation & Control Theory (AREA)
- Biomedical Technology (AREA)
- Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)
- Mathematical Optimization (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Computational Mathematics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
Abstract
The invention provides a method for determining the additional mass and the additional moment of inertia of a ship based on a neural network, which comprises the steps of firstly, solving the additional mass of the same test ship by utilizing different methods, and determining the parameter interval and the experiment group number of the additional mass; then establishing a motion model of the test ship and obtaining an objective function and constraint conditions; then constructing a simulation model for testing the ship dynamic positioning system in MATLAB, testing according to the determined number of groups of the weight matrix, and obtaining an optimal parameter scheme; and repeating the process to calculate the optimal parameter schemes of a plurality of different ship types as training samples of the neural network, and calculating the additional mass and the additional moment of inertia of any ship by using the obtained basic probability distribution. The invention has self-learning capability, better universality and generalization, and can meet the requirement of determining the additional mass of different types of ships.
Description
Technical Field
The invention relates to the field of ships, in particular to a method for determining additional mass and additional moment of inertia of any ship by using simulation test data of various ship types as training samples of a neural network.
Background
One of the key problems in accurately predicting the motion performance of a ship in waves is the accurate determination of the additional mass and damping coefficient of the ship. The additional mass during movement of the vessel is closely related to damping and fluid flow characteristics around the vessel. At present, many theoretical methods for predicting the motion performance of a ship in waves exist, including a slice theory, an elongated body theory, a three-dimensional frequency domain theory, various improved methods thereof and the like. These theoretical methods are basically built based on linear potential flow theory, but due to the limitation of the linear potential flow theory, it is difficult to consider the effects of viscosity and nonlinearity, and if no correction is applied, the prediction accuracy of the ship motion is poor.
At present, the experimental method of the additional mass and damping of the ship is mainly divided into: ① The free damping experiment of the ship swing only shows the additional mass and damping of the ship at the resonance frequency, and the frequency correlation is lacking; ② The forced oscillation motion experiment of the ship can give additional mass and damping at a plurality of frequencies, but the analysis of experimental results is still limited by subsamples, and in addition, the forced oscillation motion of the ship model has higher requirements on experimental equipment and test systems.
The additional quality obtained by the methods has errors with different degrees, but the standard method is lack as a reference, so that the accuracy of the additional quality value obtained by the methods is difficult to evaluate, and the optimal additional quality cannot be selected as a parameter for controlling the ship motion.
Disclosure of Invention
The invention aims to provide a method for determining the additional mass and the additional moment of inertia of any ship by using simulation test data of various ship types as training samples of a neural network.
Specifically, the invention provides a method for determining the additional mass and the additional moment of inertia of a ship based on a neural network, which comprises the following steps:
Step 100, solving the additional mass and the additional moment of inertia of the same test ship by using different methods, and determining the parameter interval and the experiment group number of the additional mass and the additional moment of inertia according to the additional mass and the additional moment of inertia;
step 200, firstly, a motion model of a test ship is established, then the motion model is converted into a linear form, discretization processing is carried out, the motion of the test ship is optimized, an objective function is obtained, and constraint conditions are established according to propeller constraint configuration of the test ship;
step 300, constructing a simulation model for testing the ship dynamic positioning system in MATLAB according to the motion model, the objective function and the constraint condition, testing the simulation model in a corresponding number according to the number of groups of the determined weight matrix so as to calculate the deviation between the actual path and the planned path in each group of tests, and then selecting the optimal parameter scheme of the tested ship;
Step 400, repeating steps 100-300, calculating optimal parameter schemes of a plurality of ships with different ship types, taking all the optimal parameter schemes as training samples of the neural network, obtaining a basic probability distribution value of a target element, and calculating the additional mass and the additional moment of inertia of any ship based on the basic probability distribution value.
In one embodiment of the present invention, the different methods in step 100 include regression equations, oscillation experiments, and geometric methods.
In one embodiment of the present invention, the regression equation is formulated to find the additional mass as follows:
The formula for solving the additional moment of inertia by the regression formula method is as follows:
Wherein m is the mass of the ship to be measured; m x、my is the additional mass of heave, heave and J z is the additional moment of inertia.
In one embodiment of the present invention, the step of determining the additional mass value by the oscillation test method is as follows:
step 110, placing a ship model with the mass of m in a water tank, connecting the front end of the ship model with a flat spring with the elastic coefficient of C through a connecting rod, and rigidly fixing the rear end of the ship model on a test bed;
Step 111, making the ship model-connecting rod perform longitudinal undamped oscillation, and determining an oscillation period calculation formula according to the relation between the oscillation circle frequency of the undamped oscillation coefficient, the elastic coefficient C and the mass m;
Step 112, setting the mass of the connecting rod as m 0 and the measured oscillation period as T 0; the mass of the ship model is m 0 +m after the ship model is connected with the connecting rod, and the measured oscillation period is T 1 in an air oscillation test; carrying out an oscillation test on the ship model and the connecting rod in water, wherein the total mass of the ship model and the connecting rod is m 0+m+mx, and the measured oscillation period is T 2;
Step 113, according to the oscillation period calculation formula, the additional mass m x of the ship model can be obtained by eliminating the mass m 0 and the elastic coefficient C of the connecting rod; repeating the steps 110-113, and only changing the ship model-connecting rod to do transverse undamped oscillation, thereby obtaining the additional mass m y; steps 110-113 are repeated, and only the ship model-connecting rod is changed to perform bow-sway undamped oscillation, so that an additional moment of inertia J Z can be obtained.
In one embodiment of the invention, the geometric method finds the additional mass and the additional moment of inertia as follows:
Step 120, regarding the test ship as an elliptical revolving body approximately, taking the ship length as the long axis and the draft as the short axis, and calculating the corrected additional mass of the test ship according to the difference between the ship shape and the elliptical revolving body through a sub Ke Bi correction formula;
Step 121, correcting the additional moment of inertia by using a sub-coxib correction formula according to the difference between the hull shape and the elliptic revolution body caused by the head-tail asymmetry, and finally obtaining the corrected additional moment of inertia.
In one embodiment of the present invention, in the step 200, the process of creating the motion model is as follows:
Firstly, a nonlinear ship motion model is established, and the model is expressed as follows:
Converting the nonlinear motion model into a form of a vector function:
Wherein M represents a system inertia matrix, D represents a hydrodynamic linear damping coefficient matrix, v represents a velocity matrix, Representing acceleration matrix, τ representing control force, ω representing matrix caused by environmental disturbance force, converting the established motion model intoCan be expressed as:
expanding the established motion model at a reference state point into a linear form:
Where v r is the desired navigational speed and τ r is the desired thrust.
In one embodiment of the present invention, in the step 200, the discretization is implemented by using a Forward-Euler method, and the specific process is as follows:
Let T be a sampling period, the rate of change of the acceleration deviation at time k can be obtained as follows:
substituting formula (19) into formula (I) can be achieved
Simplifying it into:
Let the predicted time domain of the control system be N P, the control time domain be N c, the state of the control system at the future time is expressed in the form of a matrix:
Wherein Y k represents a motion state matrix of the ship at a future time, Representing a velocity deviation matrix, τ (k) represents a control force matrix.
In one embodiment of the present invention, the constraint conditions are directed to a control amount limit constraint and a control increment constraint in the control process, and correspond to a thrust force and thrust moment constraint and a constraint of a thrust force and thrust moment change rate, and are expressed by the following formulas:
in one embodiment of the present invention, in the step 300, the process of calculating the deviation between the actual path and the planned path in each set of tests is as follows:
the deviation is first quantized, and the quantization parameters include:
(1) Correlation: the evaluation of the correlation is quantified by adopting a correlation coefficient R;
Where x i denotes the displacement of the actual motion trajectory in one of the three degrees of freedom of heave, heave and yaw, y i denotes the displacement of the planned trajectory in that degree of freedom, Representing the displacement average of the actual motion trail,/>Representing the displacement average value of the planned track on the degree of freedom, wherein n represents the number of data in simulation time, namely the number of cycles;
(2) Standard deviation S of deviation;
(3) Mean value sigma of the deviations;
(4) Maximum value of deviation: sigma max:
And carrying out quantization judgment on the consistency between the actual track and the target track of the ship under different additional masses and additional moments of inertia through the obtained quantization parameters, and selecting a correlation coefficient R, a standard deviation S of deviation, an average deviation sigma and the corresponding additional mass and additional moment of inertia when the average deviation sigma is minimum, thereby obtaining the optimal additional mass scheme.
In one embodiment of the present invention, in the step 400, the formula for calculating the additional mass and the additional moment of inertia of any vessel is as follows:
Mx=A1X·mx1+A2X·mx2+A3X·mx3 (29)
My=B1Y·my1+B2Y·my3+B3X·my3 (30)
jz=c1z*jz1+c2z*jz2+c3z*jz3 (31)
where Mx and My are the final determined additional masses and Jz is the additional moment of inertia.
According to the invention, simulation test data under various ship-shaped optimal parameter schemes are obtained and used as training samples of the neural network, the basic probability distribution value representing any ship is obtained after the training samples are processed by the neural network method, and the additional mass and the additional moment of inertia with the minimum error rate of any ship can be obtained through the basic probability distribution value. The invention also has self-learning capability, better universality and generalization, and can meet the requirements of different types of ships for determining the additional quality coefficient.
Drawings
FIG. 1 is a flow chart of an authentication method according to one embodiment of the present invention;
FIG. 2 is a schematic flow chart of calculating additional mass values using an oscillation test method;
FIG. 3 is a flow chart for geometrically solving for additional quality values;
FIG. 4 is a graph of additional mass and additional moment of inertia coefficients of an elliptical rotor;
FIG. 5 is a flow diagram of MATLAB modeling in accordance with one embodiment of the present invention;
FIG. 6 is a graph showing the output effect of the simulation test according to one embodiment of the present invention.
Detailed Description
In recent years, with rapid advances in computer technology and computing technology, computational Fluid Dynamics (CFD) has also been developed. The numerical simulation of the ship water movement based on the CFD theory has the advantages that the cost is low, the non-contact flow field measurement is realized, the scale effect is avoided, the influence of factors such as sensor size and model deformation on the flow field in a physical model experiment can be eliminated, more detailed flow field information can be obtained, and the like, so that the method fully utilizes the CFD to correct the calculation errors of the additional mass and the additional moment of inertia of the ship.
Specific structures and implementation procedures of the present solution are described in detail below through specific embodiments and drawings.
As shown in fig. 1, in one embodiment of the present invention, a method for determining an additional mass and an additional moment of inertia of a ship based on a neural network is disclosed, comprising the steps of:
step 100, solving the additional mass value of the same test ship by using different methods, and determining the parameter interval and the experiment group number of the additional mass according to the additional mass value;
the different methods selected in this embodiment include regression equations, oscillation tests, and geometric methods.
(1) The formula for solving the additional mass by the regression formula method is as follows:
the regression equation is used to find the additional moment of inertia as follows:
Wherein m is the mass of the ship to be measured; m x、my is the additional mass of heave, heave and J z is the additional moment of inertia.
(2) The oscillation test method comprises the following steps of:
step 110, placing a ship model with the mass of m in a water tank, connecting the front end of the ship model with a flat spring with an elastic system of C through a connecting rod, and rigidly fixing the rear end of the ship model on a test bed;
the connecting rod here denotes both a connecting rod and a device for driving the connecting rod.
Step 111, making the ship model-connecting rod perform longitudinal undamped oscillation, and determining an oscillation period calculation formula according to the relation between the oscillation circle frequency of the undamped oscillation coefficient, the elastic coefficient C and the mass m;
Wherein, the relation between the oscillation circle frequency of the undamped oscillation coefficient, the elastic coefficient C and the mass m is as follows:
the oscillation period calculation formula is as follows:
Step 112, setting the mass of the connecting rod as m 0, and determining the oscillation period as T 0; the mass of the ship model is m 0 +m after the ship model is connected with the connecting rod, and the measured oscillation period is T 1 in an air oscillation test; carrying out an oscillation test on the ship model and the connecting rod in water, wherein the total mass of the ship model and the connecting rod is m 0+m+mx, and the measured oscillation period is T 2;
in the steps: t 0、T1、T2 has the following calculation formula:
Step 113, according to the oscillation period calculation formula, the additional mass m x of the ship model can be obtained by eliminating the mass m 0 and the elastic coefficient C of the connecting rod; repeating the steps 110-113, and only changing the ship model-connecting rod to do transverse undamped oscillation, thereby obtaining the additional mass m y; steps 110-113 are repeated, and only the ship model-connecting rod is changed to do transverse undamped oscillation, so that the additional moment of inertia J Z can be obtained.
The additional mass calculation formula of m x in this step is:
the obtained m y additional mass calculation formula is as follows:
the obtained m y additional mass calculation formula is as follows:
(3) The geometric method finds the formulas of the additional mass and the additional moment of inertia as follows:
Step 120, regarding the test ship as an elliptical revolving body approximately, taking the ship length as the long axis and the draft as the short axis, and calculating the corrected additional mass of the test ship according to the difference between the ship shape and the elliptical revolving body through a sub Ke Bi correction formula;
The calculation formula is as follows:
mx=kxm (12)
Where k x、ky and k z are determined by fig. 4, and the major axis is 2a and the minor axis is 2b as determined in fig. 4.
Step 121, correcting the additional moment of inertia by using a sub-coxib correction formula according to the difference between the hull shape and the elliptic revolution body caused by the head-tail asymmetry, and finally obtaining the corrected additional moment of inertia.
The subfraction Ke Bi correction formula for the additional moment of inertia is as follows:
(4) The parameter intervals and experimental groups were determined as follows:
Three additional masses were determined by the three methods described above, as follows:
Method of | Additional mass |
Regression formula | (mx1,my1) |
Oscillation test | (mx2,my2) |
Geometric method | (mx3,my3) |
Taking the minimum value and the maximum value of three groups of additional masses m x and the minimum value and the maximum value of m y, and determining the intervals of the parameter scheme as [ m xmin,mxmax ] and [ m ymin,mymax ]; the experimental groups were subdivided, and the parameters of each experimental group were shown in the following table, assuming that the number of groups of experimental groups was X.
Experimental group | Experiment group 1 | Experiment group 2 | Experiment group 3 | Experiment group 4 | Experiment group 5 | .... |
mx | ||||||
my |
Wherein the additional mass data for each experimental group is populated according to the experimental results.
Step 200, firstly, a motion model of a test ship is established, then the motion model is converted into a linear form, discretization processing is carried out, the motion of the test ship is optimized, an objective function is obtained, and constraint conditions are established according to propeller constraint configuration of the test ship;
(1) According to the specific test ship under study, the process of establishing the motion model is as follows:
firstly, a nonlinear ship motion model is established, and the model is expressed as follows:
the nonlinear motion model is written in the form of a vector function, and since the nonlinear model of the ship motion has been converted into a linear model during the modeling process, the linear model is also used in the subsequent steps, and thus the linear motion model of the ship is directly given here.
Wherein M represents a system inertia matrix, D represents a hydrodynamic linear damping coefficient matrix, v represents a velocity matrix,Representing acceleration matrix, τ representing control force, ω representing matrix caused by environmental disturbance force, converting the established motion model intoCan be expressed as:
Expanding the established motion model at a reference state point (expected speed and heading) into a linear form:
Where v r is a desired speed, τ r is a desired thrust (in a state where the desired speed and heading are satisfied, there should be a desired thrust corresponding to the state).
(2) Discretizing by adopting a Forward-Euler method, and setting T as a sampling period to obtain the change rate of acceleration deviation at k time, wherein the change rate is as follows:
Substituting the formula (18) into the formula can be obtained
Simplifying it into:
Let the prediction time domain of the control system be N P, the control time domain be N c, the state of the system at the future time is expressed in the form of a matrix:
Wherein Y k represents a motion state matrix of the ship at a future time, Representing a velocity deviation matrix, τ (k) represents a control force matrix.
(3) The process of obtaining the objective function is as follows:
the control law is obtained through quadratic programming, and in ship motion control, the optimization targets are as follows:
1) Converging the current spatial position to a reference value as soon as possible (bringing the position and heading of the vessel close to the desired position and heading as soon as possible, even if the deviation of the position and heading approaches zero as soon as possible);
2) The current motion state is converged to a reference value as soon as possible (the speed and heading of the ship are made to approach the expected position and heading as soon as possible, even if the deviation of the speed and heading approaches zero as soon as possible);
3) The control amount is minimum (the thrust is minimum, and the deviation between the thrust and 0 can be regarded as minimum);
the deviation is generally considered as a sum of squares, so to consider the effect of each part, a weighted sum of squares can be expressed as follows:
min(η-ηr)TQ(η-ηr)+(v-vr)TR(v-vr)+τTPτ (24)
The constraint condition mainly considers the limit constraint of the control quantity and the increment constraint of the control in the control process, namely the constraint of the thrust force and the thrust moment and the constraint of the change rate of the thrust force and the thrust moment, and can be expressed as follows:
step 300, constructing a simulation model for testing the ship dynamic positioning system in MATLAB according to the motion model, the objective function and the constraint condition, testing the simulation model in a corresponding number according to the number of groups of the determined weight matrix so as to calculate the difference between the actual path and the planned path in each group of tests, and then selecting the optimal parameter scheme of the tested ship;
The constructed simulation model is shown in fig. 5, the planned path is input into the simulation model, the simulation model works, and the actual track of the ship can be output.
The effect diagram of simulation test output is shown in fig. 6, wherein the diagram (a) is the comparison of a planned path and a simulated track in the heave direction, a solid line represents the planned path, and a dotted line represents the actual track; the graph (b) is the comparison of the planned path and the simulated track in the sway direction, the solid line represents the planned path, and the dotted line represents the actual track; the graph (c) is a comparison of the planned path and the simulated trajectory in the yaw direction, the solid line represents the planned path, and the broken line represents the actual trajectory.
The process of selecting the optimal parameter scheme is as follows:
correlation: the evaluation of the correlation is quantified by adopting a correlation coefficient R;
Where x i represents the displacement of the actual motion profile in one of the three degrees of freedom of heave, heave and yaw, y i represents the displacement of the planned trajectory in that degree of freedom, x represents the displacement average of the actual motion profile, The displacement average value of the planned track in the degree of freedom is represented, and n represents the number of data in the simulation time, namely the number of loops.
Standard deviation S of deviation;
Average deviation sigma;
maximum value of deviation: σ max.
The obtained three quantization parameters can be used for carrying out quantization judgment on the consistency between the actual track and the target track of the ship under different additional masses and additional moments of inertia, and the optimal additional mass scheme is obtained by selecting the correlation coefficient R, the standard deviation S of deviation, the average deviation sigma and the corresponding additional mass and additional moment of inertia when the average deviation sigma is minimum.
Step 400, repeating steps 100-300, calculating optimal parameter schemes of a plurality of ships with different ship types, taking all the optimal parameter schemes as training samples of the neural network, obtaining basic probability distribution of target elements, and calculating the additional mass and the additional moment of inertia of any ship based on the basic probability distribution value.
The neural network training process is as follows:
A series of samples are established by utilizing data of various ship types and are used for training a neural network, a MATLAB tool box is used for establishing a BP neural network, a logsig function is used as an activation function to limit network output to a (0, 1) interval under the condition that convergence accuracy and convergence speed are optimal, a trainlm function is used as a training function, a learnpbm function is used as a learning function, and a target error value epsilon=0.001 is taken for network training.
After training n times, the training is stable, and the network training is finished. At this time, the sample set can be input into the neural network for recognition, so as to obtain decision output of the BP neural network, and the output result is normalized so as to obtain basic probability distribution of the target element.
The additional mass and additional moment of inertia of any vessel are calculated as follows:
Mx=A1X·mx1+A2X·mx2+A3X·mx3 (29)
My=B1Y·my1+B2Y·my3+B3X·my3 (30)
jz=c1z*jz1+c2z*jz2+c3z*jz3 (31)
where Mx and My are the final determined additional mass and additional moment of inertia.
According to the method, simulation test data under various ship-shaped optimal parameter schemes are obtained and used as training samples of the neural network, the basic probability distribution value representing any ship is obtained after the training samples are processed by the neural network method, and the additional mass and the additional moment of inertia with the minimum error rate of any ship can be obtained through the basic probability distribution value. The invention also has self-learning capability, better universality and generalization, and can meet the requirements of different types of ships for determining the additional quality coefficient.
By now it should be appreciated by those skilled in the art that while a number of exemplary embodiments of the invention have been shown and described herein in detail, many other variations or modifications of the invention consistent with the principles of the invention may be directly ascertained or inferred from the present disclosure without departing from the spirit and scope of the invention. Accordingly, the scope of the present invention should be understood and deemed to cover all such other variations or modifications.
Claims (3)
1. The method for determining the additional mass and the additional moment of inertia of the ship based on the neural network is characterized by comprising the following steps of:
step 100, solving the additional mass and the additional moment of inertia of the same test ship by using different methods, and determining the parameter interval and the experiment group number of the additional mass and the additional moment of inertia according to the additional mass and the additional moment of inertia; the different methods comprise a regression formula method, an oscillation test method and a geometric method;
step 200, firstly, a motion model of a test ship is established, then the motion model is converted into a linear form, discretization processing is carried out, the motion of the test ship is optimized, an objective function is obtained, and constraint conditions are established according to propeller constraint configuration of the test ship;
step 300, constructing a simulation model for testing the ship dynamic positioning system in MATLAB according to the motion model, the objective function and the constraint condition, testing the simulation model in a corresponding number according to the number of groups of the determined weight matrix so as to calculate the deviation between the actual path and the planned path in each group of tests, and then selecting the optimal parameter scheme of the tested ship;
Step 400, repeating the steps 100-300, calculating optimal parameter schemes of a plurality of ships with different ship types, taking all the optimal parameter schemes as training samples of a neural network, obtaining a basic probability distribution value of a target element, and calculating the additional mass and the additional moment of inertia of any ship based on the basic probability distribution value; the formulas for calculating the additional mass and the additional moment of inertia of any vessel are as follows:
;
where Mx and My are the additional mass that is ultimately determined, jz is the additional moment of inertia that is ultimately determined, Representing the additional mass value in the sway direction calculated using the regression equation; /(I)Representing the additional mass value in the sway direction calculated using the oscillation test method; /(I)Representing the calculated additional mass value in the sway direction by using the geometric method; /(I)Representing the additional mass value in the heave direction calculated using the regression equation; /(I)Representing the additional mass value in the heave direction calculated by the oscillation test method; /(I)Representing the additional mass value in the heave direction calculated using the geometrical method.
2. The method for determining according to claim 1, wherein,
The step of obtaining the additional mass value by the oscillation test method is as follows:
step 110, placing a ship model with the mass of m in a water tank, connecting the front end of the ship model with a flat spring with the elastic coefficient of C through a connecting rod, and rigidly fixing the rear end of the ship model on a test bed;
Step 111, making the ship model-connecting rod perform longitudinal undamped oscillation, and determining an oscillation period calculation formula according to the relation between the oscillation circle frequency of the undamped oscillation coefficient, the elastic coefficient C and the mass m;
Step 112, setting the mass of the connecting rod as m 0 and the measured oscillation period as T 0; the mass of the ship model is m 0 +m after the ship model is connected with the connecting rod, and the measured oscillation period is T 1 in an air oscillation test; carrying out an oscillation test on the ship model and the connecting rod in water, wherein the total mass of the ship model and the connecting rod is m 0+m+mx, and the measured oscillation period is T 2;
Step 113, according to the oscillation period calculation formula, the additional mass m x of the ship model can be obtained by eliminating the mass m 0 and the elastic coefficient C of the connecting rod; repeating the steps 110-113, and only changing the ship model-connecting rod to do transverse undamped oscillation, thereby obtaining the additional mass m y; repeating steps 110-113, and changing ship model-connecting rod to make bow-shake undamped oscillation only so as to obtain additional moment of inertia 。
3. The method for determining according to claim 1, wherein,
The steps of the geometric method for obtaining the additional mass and the additional moment of inertia are as follows:
Step 120, regarding the test ship as an elliptical revolving body approximately, taking the ship length as the long axis and the draft as the short axis, and calculating the corrected additional mass of the test ship according to the difference between the ship shape and the elliptical revolving body through a sub Ke Bi correction formula;
Step 121, correcting the additional moment of inertia by using a sub-coxib correction formula according to the difference between the hull shape and the elliptic revolution body caused by the head-tail asymmetry, and finally obtaining the corrected additional moment of inertia.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010800265.5A CN112084574B (en) | 2020-08-11 | 2020-08-11 | Ship additional mass and additional moment of inertia determining method based on neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010800265.5A CN112084574B (en) | 2020-08-11 | 2020-08-11 | Ship additional mass and additional moment of inertia determining method based on neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112084574A CN112084574A (en) | 2020-12-15 |
CN112084574B true CN112084574B (en) | 2024-04-19 |
Family
ID=73734937
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010800265.5A Active CN112084574B (en) | 2020-08-11 | 2020-08-11 | Ship additional mass and additional moment of inertia determining method based on neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112084574B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859147A (en) * | 2010-06-11 | 2010-10-13 | 哈尔滨工程大学 | Ship course intelligent coordination control method |
CN108846207A (en) * | 2018-06-15 | 2018-11-20 | 哈尔滨工程大学 | A kind of ship movement prediction device and its forecasting procedure based on Non-linear Model Identification |
KR20180126730A (en) * | 2017-05-18 | 2018-11-28 | (주)세이프텍리서치 | Autonomous navigation system using ship dynamics model |
CN109656143A (en) * | 2019-02-25 | 2019-04-19 | 曲阜师范大学 | A kind of sea drives the adaptive tracking control method and system of ship entirely |
CN110647041A (en) * | 2019-10-11 | 2020-01-03 | 大连海事大学 | Unmanned ship model full-coefficient accurate identification method |
CN110737986A (en) * | 2019-10-15 | 2020-01-31 | 大连海事大学 | unmanned ship energy efficiency intelligent optimization simulation system and method |
-
2020
- 2020-08-11 CN CN202010800265.5A patent/CN112084574B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101859147A (en) * | 2010-06-11 | 2010-10-13 | 哈尔滨工程大学 | Ship course intelligent coordination control method |
KR20180126730A (en) * | 2017-05-18 | 2018-11-28 | (주)세이프텍리서치 | Autonomous navigation system using ship dynamics model |
CN108846207A (en) * | 2018-06-15 | 2018-11-20 | 哈尔滨工程大学 | A kind of ship movement prediction device and its forecasting procedure based on Non-linear Model Identification |
CN109656143A (en) * | 2019-02-25 | 2019-04-19 | 曲阜师范大学 | A kind of sea drives the adaptive tracking control method and system of ship entirely |
CN110647041A (en) * | 2019-10-11 | 2020-01-03 | 大连海事大学 | Unmanned ship model full-coefficient accurate identification method |
CN110737986A (en) * | 2019-10-15 | 2020-01-31 | 大连海事大学 | unmanned ship energy efficiency intelligent optimization simulation system and method |
Non-Patent Citations (2)
Title |
---|
A robust neural controller for underwater robot manipulators;Minho Lee, Hyeung-Sik Choi;IEEE Transactions on Neural Networks;第11卷(第6期);全文 * |
基于RBF神经网络的船舶路径跟踪控制建模及仿真;仇明;张贺;金博文;曾骥;;海洋工程装备与技术(第05期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN112084574A (en) | 2020-12-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108121856B (en) | Dynamic stability analysis method for full-flight-domain aircraft | |
CN106896722B (en) | The hypersonic vehicle composite control method of adoption status feedback and neural network | |
Chauhan et al. | Review of aerodynamic parameter estimation techniques | |
CN115496006A (en) | High-precision numerical simulation method suitable for hypersonic aircraft | |
CN116150934A (en) | Ship maneuvering Gaussian process regression online non-parameter identification modeling method | |
CN109857127B (en) | Method and device for calculating training neural network model and aircraft attitude | |
CN112084574B (en) | Ship additional mass and additional moment of inertia determining method based on neural network | |
CN104899642B (en) | Large deformation flexible body dynamic stress compensation method based on hybrid production style | |
Deng et al. | A coupling method of geomagnetic aided inertial attitude errors | |
CN114030008B (en) | Industrial robot practical training energy consumption measurement method based on data driving | |
Milovanović et al. | Adaptive control of nonlinear MIMO system with orthogonal endocrine intelligent controller | |
CN114462149B (en) | Aircraft pneumatic parameter identification method based on pre-training and incremental learning | |
CN113158297B (en) | Hydraulic arc steel gate parameter load identification method | |
CN112327666B (en) | Method for determining target function weight matrix of power cruise system control model | |
CN114186477A (en) | Elman neural network-based orbit prediction algorithm | |
Yuqi et al. | Time-varying parameters estimation with adaptive neural network EKF for missile-dual control system | |
Ghosh Roy et al. | Longitudinal aircraft parameter estimation using neuro-fuzzy and genetic algorithm based method | |
Gubarev et al. | Identification in cognitive maps in the impulse processes mode with full information | |
Petrich | Improved guidance, navigation, and control for autonomous underwater vehicles: theory and experiment | |
Simsek et al. | System Identification and Handling Quality Analysis of a UAV from Flight Test Data | |
CN116011350B (en) | Deep reinforcement learning black box identification method for module ship motion characteristic analysis | |
Kimathi et al. | Robust Aerodynamic Parameter Estimation of Unmanned Aircraft Based on Two-step Identification | |
CN117875090B (en) | Fixed-wing unmanned aerial vehicle incremental element flight aerodynamic modeling method considering wind interference | |
CN117685981A (en) | Method and system for predicting 4D flight path of aircraft based on equivalent guidance law | |
Ji et al. | Unsteady aerodynamic prediction using limited samples based on transfer learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |