CN115859485A - Streamline seed point selection method based on ship appearance characteristics - Google Patents

Streamline seed point selection method based on ship appearance characteristics Download PDF

Info

Publication number
CN115859485A
CN115859485A CN202310165021.8A CN202310165021A CN115859485A CN 115859485 A CN115859485 A CN 115859485A CN 202310165021 A CN202310165021 A CN 202310165021A CN 115859485 A CN115859485 A CN 115859485A
Authority
CN
China
Prior art keywords
boundary line
line
area
points
ship
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.)
Granted
Application number
CN202310165021.8A
Other languages
Chinese (zh)
Other versions
CN115859485B (en
Inventor
唐滨
刘昊康
王昊东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Qingdao Harbin Engineering University Innovation Development Center
Original Assignee
Qingdao Harbin Engineering University Innovation Development Center
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Qingdao Harbin Engineering University Innovation Development Center filed Critical Qingdao Harbin Engineering University Innovation Development Center
Priority to CN202310165021.8A priority Critical patent/CN115859485B/en
Publication of CN115859485A publication Critical patent/CN115859485A/en
Application granted granted Critical
Publication of CN115859485B publication Critical patent/CN115859485B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a streamline seed point selection method based on ship appearance characteristics, and belongs to the field of scientific calculation visualization. The method comprises the following steps of collecting and constructing a sample data set, constructing a BP (back propagation) neural network, predicting all points on a next layer boundary line from all points of a longitudinal and middle section contour line or all points on one region boundary line, respectively inputting constructed test set data into the BP neural network for training, constructing a first region boundary line network model, a second region boundary line network model, a third region boundary line network model and a fourth region boundary line network model, and respectively inputting test set data into the trained boundary line network models to obtain corresponding boundary lines. By constructing a streamline seed point distribution area boundary line screening model and dividing a streamline seed point distribution area, the flow field characteristics can be accurately reflected through the streamline on the premise that fewer seed points are selected and fewer streamlines are generated.

Description

Streamline seed point selection method based on ship appearance characteristics
Technical Field
The invention relates to a streamline seed point selection method based on ship appearance characteristics, and belongs to the technical field of scientific calculation visualization.
Background
The generation of the streamline requires to know the velocity vector of each point in the flow field, then select a seed point, and start integration from the seed point according to the positive direction or the negative direction of the velocity at the point. Before the integration, an appropriate integration step size is set, such as a distance for each step of integration is specified. That is, the streamline starts from the seed point position, integrates with a proper step length along the velocity vector of the seed point position, and extends to the positive direction or the negative direction, so as to obtain the complete streamline.
In the streamline generation process, the selected positions of the seed points influence the effect of streamline distribution, theoretically, one seed point can generate a streamline through calculation, and the more densely the seed points are distributed, the more densely the streamline is distributed.
Taking the generation of a basin streamline near a ship as an example, the manual selection of streamline seed points in actual engineering follows a principle: in areas with large deformation, such as a bow and a stern, the streamlines are distributed densely in areas close to the ship. The invention also provides a streamline seed point selection method based on ship appearance characteristics, which can automatically divide a flow field area of a river basin near a ship into a characteristic area needing to select more kinds of seed points and a non-characteristic area needing to select a proper amount of seed points.
In order to better meet the actual requirements, in the process of specifically realizing the method, not only a characteristic region and a non-characteristic region are divided, but also the region is divided in multiple levels by combining factors such as the distance from the surface of the ship, the angle of water flow when impacting the ship and the like, so that the flow field characteristics of a watershed near the ship can be better displayed.
Disclosure of Invention
The invention aims to provide a streamline seed point selection method based on ship appearance characteristics.
In order to achieve the purpose, the invention is realized by the following technical scheme:
step 1: and collecting and constructing a sample data set.
Step 1-1: selecting a ship model to obtain a longitudinal section of the ship model, selecting a first area boundary line, a second area boundary line, a third area boundary line and a fourth area boundary line, dividing an area between a ship longitudinal section profile line and the first boundary line into a first area, dividing an area between the first boundary line and the second boundary line into a second area, dividing an area between the second boundary line and the third boundary line into a third area, and dividing an area between the third boundary line and the fourth boundary line into a fourth area.
Step 1-2: the vertical middle section contour line is taken as a boundary line 0, the first zone boundary line is taken as a boundary line 1, the second zone boundary line is taken as a boundary line 2, the third zone boundary line is taken as a boundary line 3, and the fourth zone boundary line is taken as a boundary line 4.
Step 1-3: constructing a first sample data set by taking the boundary line 0 as an input and the boundary line 1 as an output; constructing a second sample data set by taking the boundary line 1 as input, the boundary line 2 as output, the boundary line 2 as input and the boundary line 3 as output; constructing a fourth sample data set by taking the boundary line 3 as input and the boundary line 4 as output; and divided into a test set and a training set.
Step 2: constructing a BP neural network, and predicting all points on a boundary line of a next layer from all points of a profile line of a longitudinal-medial section or all points on a boundary line of a region; the BP neural network hidden layer has 3 layers, the number of input and output neurons is m, and m is equal to the number of points taken on the boundary line.
And 3, step 3: and respectively inputting the constructed test set data into BP neural network training to construct a first area boundary line network model, a second area boundary line network model, a third area boundary line network model and a fourth area boundary line network model.
And 4, step 4: and respectively inputting the test set data into the trained boundary line network model to obtain corresponding boundary lines.
Preferably, the selecting step of the first area boundary line is as follows:
the method comprises the steps of collecting streamline seed points at set intervals according to the longitudinal and middle profile contour line of a ship in the total length of the ship, wherein the total length of the ship is the maximum horizontal distance between the head end and the tail end of the surface of the ship, and the maximum horizontal distance is parallel to a design waterline.
And calculating the minimum included angle between a connecting line between two adjacent points of each point on the profile line of the longitudinal-middle section and a horizontal line, and calculating the slope of the vertical line of the connecting line between two adjacent points.
According to the slope of the vertical line obtained by the connecting line of the two adjacent points of the point, the slope extends a distance from the point to the direction far away from the profile contour line of the ship; connecting all end points of the extended vertical line to form a first area boundary line; and the area between the first area boundary line and the longitudinal and middle section contour lines is a first area.
Preferably, the selecting steps of the second area boundary line, the third area boundary line and the fourth area boundary line are as follows:
acquiring streamline seed points at a set distance on the boundary line of the previous area according to the maximum horizontal distance between the head end point and the tail end point of the previous area; said spacing is
Figure SMS_1
In which>
Figure SMS_2
Representing the overall length of the vessel.
And calculating the minimum included angle between a connecting line between two adjacent points of each point on the boundary line of the previous area and the horizontal line, and calculating the slope of the perpendicular line of the connecting line between two adjacent points.
According to the slope of the vertical line obtained by the connecting line of the two adjacent points of the point, the slope extends a distance from the point to the direction far away from the middle section contour line of the ship; connecting all the end points of the extended vertical line to form a boundary line of the current area; and the area between the current area boundary line and the previous area boundary line is the current area.
Preferably, the vertical line is extended as follows:
if the minimum angle between the line connecting two adjacent points of the point and the horizontal line is within the interval of 0 degrees and 30 degrees, the perpendicular line connecting the adjacent points through the point extends outwards by the distance of t.
If the minimum angle between the line connecting two adjacent points of the point and the horizontal line is within the interval of (30 DEG, 60 DEG), the line connecting the adjacent points through the point is extended outward by a distance of 1.25 x t in the perpendicular direction.
If the minimum angle between the line connecting two adjacent points of the point and the horizontal line is within the interval (60 DEG, 90 DEG), the line connecting the adjacent points is extended outward by a distance of 1.5 x t through the point.
When each point of the first area boundary line is selected, T =0.05 × T, when the second area boundary line is selected, T =0.1 × T, when the third area boundary line is selected, T =0.15 × T, and when the fourth area boundary line is selected, T =0.2 × T, wherein T represents the draught depth of the ship.
Preferably, the specific steps of step 3 are as follows:
step 3-1: initializing BP neural network parameters, and setting learning rate of BP neural network as
Figure SMS_5
，/>
Figure SMS_8
Is->
Figure SMS_11
The weight matrix of the BP neural network is compared with>
Figure SMS_4
And a bias matrix pick>
Figure SMS_9
Is set to->
Figure SMS_12
In between, a random matrix, is selected>
Figure SMS_14
Is the BP neural network->
Figure SMS_3
The fifth on the hidden layer>
Figure SMS_7
Weights on individual neurons, based on the neural signal>
Figure SMS_10
Is the number ^ greater or lesser of the bp neural network>
Figure SMS_13
The fifth on the hidden layer>
Figure SMS_6
Bias values on individual neurons.
Step 3-2: inputting the constructed training set into a BP neural network, iterating the sample data in the BP neural network for P times, and obtaining the MSE function with the minimum error value as
Figure SMS_16
Will >>
Figure SMS_19
And a target value->
Figure SMS_22
By contrast, if->
Figure SMS_17
Less than the target value>
Figure SMS_18
Then the iteration is stopped, if->
Figure SMS_21
Greater than the target value>
Figure SMS_23
Then the iteration continues until pick>
Figure SMS_15
Less than target value
Figure SMS_20
。
Preferably, the weight matrix
Figure SMS_24
；
Figure SMS_25
Is a weight matrix->
Figure SMS_26
Is/are>
Figure SMS_27
A sub-status value, <' > or>
Figure SMS_28
Is->
Figure SMS_29
The first derivative of (a).
Bias matrix
Figure SMS_30
；
In the formula (I), the compound is shown in the specification,
Figure SMS_31
is a bias matrix>
Figure SMS_32
Is/are>
Figure SMS_33
Sub-status value>
Figure SMS_34
Is->
Figure SMS_35
The first derivative of (a).
Preferably, the activation function of the BP neural network adopts a sigmoid function, and the mean square error function of the loss function adopts an MSE function.
The sigmoid function is:
Figure SMS_36
；
in the formula (I), the compound is shown in the specification,
Figure SMS_37
represents the value of the neuron after calculation by the activation function, is/are>
Figure SMS_38
Representing the jth input value of the ith hidden layer.
The MSE function is:
Figure SMS_39
；
in the formula (I), the compound is shown in the specification,mis the number of points on the boundary line,
Figure SMS_40
is the coordinate of the ith point of the set of points on the current solution boundary line c, and>
Figure SMS_41
the coordinates of the ith point in the set of points on the current solved boundary line c are shown, the value of c is 1,2,3 and 4,
Figure SMS_42
is the variance value of the two point coordinate values, i.e.>
Figure SMS_43
The value of (c).
Preferably, the composition further comprises, among others,
Figure SMS_44
；
in the formula (I), the compound is shown in the specification,
Figure SMS_45
representing the total length of the ship, and c is the boundary line of the current solution.
The invention has the advantages that: the invention provides a streamline seed point distribution strategy from the aspect of ship surface characteristics, aiming at the situation that the flow field is possibly influenced by the particularity of the ship appearance to generate complex characteristics. The flow field characteristics can be accurately reflected through the flow lines on the premise of selecting fewer seed points and generating fewer flow lines.
The lightweight generation of the streamline is realized: on one hand, the number of the seed points is increased for the characteristic region, and on the premise of ensuring that the streamline effect is not influenced for the non-characteristic region, the number of the seed points is properly reduced, so that the reasonable utilization rate of computing resources is improved; on the other hand, when the number of the flow lines is reduced, researchers can capture flow field characteristics more when observing the flow lines, and sight confusion caused by huge number of the flow lines is avoided.
The automatic generation of the streamline is realized: aiming at the means of manually selecting the characteristic regions commonly used by researchers at present, the automatic selection of the characteristic regions is realized. A BP neural network model is set up, so that the seed point distribution area selection and the seed point selection are self-learning, self-organizing and self-adapting.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
FIG. 1 is a schematic view of the flow structure of the present invention.
FIG. 2 is a schematic diagram of the zone division according to the present invention.
FIG. 3 is a schematic diagram of the minimum angle between the line connecting two adjacent points and the horizontal line.
FIG. 4 is a schematic diagram of the process of the present invention for taking the point on the boundary line of the No. 1 region from the point on the profile line of the vertical section.
FIG. 5 is a schematic diagram of streamline effects under a general selection method.
Fig. 6 is a streamline schematic diagram based on the ship-type feature selection method of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
A streamline seed point selecting method based on ship appearance characteristics comprises the following steps:
step 1: and collecting and constructing a sample data set.
Step 1-1: selecting a ship model to obtain a longitudinal section of the ship model, selecting a first area boundary line, a second area boundary line, a third area boundary line and a fourth area boundary line, dividing an area between a ship longitudinal section profile line and the first boundary line into a first area, dividing an area between the first boundary line and the second boundary line into a second area, dividing an area between the second boundary line and the third boundary line into a third area, and dividing an area between the third boundary line and the fourth boundary line into a fourth area.
The selection step of the first area boundary line is as follows:
the method comprises the steps of collecting streamline seed points at set intervals according to the profile contour line of the ship overall length in the ship longitudinal and middle direction, wherein the overall length of the ship is the maximum horizontal distance which is parallel to a design waterline between the head end and the tail end of the surface of the ship.
And calculating the minimum included angle between a connecting line between two adjacent points of each point on the profile line of the longitudinal-middle section and the horizontal line, and calculating the slope of the perpendicular line of the connecting line between two adjacent points.
According to the slope of the vertical line obtained by the connecting line of the two adjacent points of the point, the slope extends a distance from the point to the direction far away from the profile contour line of the ship; connecting all end points of the extended vertical line to form a first area boundary line; and the area between the first area boundary line and the longitudinal and middle section contour lines is a first area.
The second area boundary line, the third area boundary line and the fourth area boundary line are selected as follows:
acquiring streamline seed points at a set interval on the boundary line of the previous area according to the maximum horizontal distance between the head end and the tail end of the previous area; said spacing is
Figure SMS_46
Wherein is present>
Figure SMS_47
Representing the overall length of the vessel.
And calculating the minimum included angle between a connecting line between two adjacent points of each point on the boundary line of the previous area and the horizontal line, and calculating the slope of the perpendicular line of the connecting line between two adjacent points.
According to the slope of the vertical line obtained by the connecting line of the two adjacent points of the point, the slope extends a distance from the point to the direction far away from the profile contour line of the ship; connecting all the end points of the extended vertical line to form a boundary line of the current area; and the area between the current area boundary line and the previous area boundary line is the current area.
The vertical line extending mode is as follows:
if the minimum angle between the line connecting two adjacent points of the point and the horizontal line is within the interval of 0 degrees and 30 degrees, the perpendicular line connecting the adjacent points through the point extends outwards by the distance of t.
If the minimum angle between the line connecting two adjacent points of the point and the horizontal line is within the interval (30 degrees, 60 degrees), the perpendicular line connecting the adjacent points through the point is extended outwards by a distance of 1.25 x t.
If the minimum angle between the line connecting two adjacent points of the point and the horizontal line is within the interval of (60 DEG, 90 DEG), the line connecting the adjacent points through the point is extended outward by a distance of 1.5 x t from the perpendicular line.
When each point of the first area boundary line is selected, T =0.05 × T, when the second area boundary line is selected, T =0.1 × T, when the third area boundary line is selected, T =0.15 × T, and when the fourth area boundary line is selected, T =0.2 × T, wherein T represents the draught depth of the ship.
According to the effect of the region division, the region areas are sorted from small to large as follows: region 1, region 2, region 3, and region 4. On the premise that the number of the seed points in the four regions is the same, the density of the seed points is sorted from small to large as follows: zone 4, zone 3, zone 2, zone 1. This effect satisfies the expression "the closer the area to the surface of the vessel, the denser the distribution of the seed points".
Step 1-2: the vertical middle section contour line is taken as a boundary line 0, the first zone boundary line is taken as a boundary line 1, the second zone boundary line is taken as a boundary line 2, the third zone boundary line is taken as a boundary line 3, and the fourth zone boundary line is taken as a boundary line 4.
Step 1-3: constructing a first sample data set by taking the boundary line 0 as an input and the boundary line 1 as an output; constructing a second sample data set by taking the boundary line 1 as input, the boundary line 2 as output, the boundary line 2 as input and the boundary line 3 as output; constructing a fourth sample data set by taking the boundary line 3 as an input and the boundary line 4 as an output; and divided into a test set and a training set.
And 2, step: constructing a BP neural network, and predicting all points on a boundary line of a next layer from all points of a profile line of a longitudinal-medial section or all points on a boundary line of a region; the BP neural network hidden layer has 3 layers, the number of input and output neurons is m, and m is equal to the number of streamline seed points on the boundary line.
And 3, step 3: and respectively inputting the constructed test set data into BP neural network training to construct a first area boundary line network model, a second area boundary line network model, a third area boundary line network model and a fourth area boundary line network model.
The specific steps of the step 3 are as follows:
step 3-1: initializing BP neural network parameters and setting BP neural network learning rate as
Figure SMS_49
，/>
Figure SMS_54
Is->
Figure SMS_57
The weight matrix of the BP neural network is compared with>
Figure SMS_50
And a bias matrix pick>
Figure SMS_52
Is set to->
Figure SMS_55
In between, a random matrix, is selected>
Figure SMS_58
For BP neural network ^ th->
Figure SMS_48
The fifth on the hidden layer>
Figure SMS_53
Weight on each neuron->
Figure SMS_56
Is the number ^ greater or lesser of the bp neural network>
Figure SMS_59
The fifth on the hidden layer>
Figure SMS_51
Bias values on individual neurons.
The weight matrix
Figure SMS_60
；
Figure SMS_61
Is a weight matrix->
Figure SMS_62
Is/are>
Figure SMS_63
A sub-status value, <' > or>
Figure SMS_64
Is->
Figure SMS_65
The first derivative of (a).
Bias matrix
Figure SMS_66
；
In the formula (I), the compound is shown in the specification,
Figure SMS_67
is a bias matrix->
Figure SMS_68
Is/are>
Figure SMS_69
A sub-status value, <' > or>
Figure SMS_70
Is->
Figure SMS_71
The first derivative of (a). />
Step 3-2: inputting the constructed training set into a BP neural network, iterating P times of sample data in the BP neural network, and obtaining the minimum value of the error of the MSE function as
Figure SMS_73
Will >>
Figure SMS_76
And a target value->
Figure SMS_78
By contrast, if->
Figure SMS_74
Less than the target value>
Figure SMS_77
Then the iteration is stopped, if->
Figure SMS_79
Greater than the target value>
Figure SMS_80
Then the iteration continues until pick>
Figure SMS_72
Less than target value
Figure SMS_75
。
Wherein, the first and the second end of the pipe are connected with each other,
Figure SMS_81
；
in the formula (I), the compound is shown in the specification,
Figure SMS_82
representing the total length of the ship, and c is the boundary line of the current solution.
The activation function of the BP neural network adopts a sigmoid function, and the mean square error function of the loss function adopts an MSE function.
The sigmoid function is:
Figure SMS_83
；
in the formula (I), the compound is shown in the specification,
Figure SMS_84
represents the value of the neuron after calculation by the activation function, is/are>
Figure SMS_85
Representing the jth input value of the ith hidden layer.
The MSE function is:
Figure SMS_86
；
in the formula (I), the compound is shown in the specification,mis the number of points on the boundary line,
Figure SMS_87
is the coordinate of the ith point of the set of points on the current solution boundary line c, < >>
Figure SMS_88
Representing the coordinates of the ith point in the set of points on the current solution boundary line c,the value of c is 1,2,3,4,
Figure SMS_89
is the variance value of the two point coordinate values, i.e.>
Figure SMS_90
The value of (c).
And 4, step 4: and respectively inputting the test set data into the trained boundary line network model to obtain corresponding boundary lines.
In a flow field area of a watershed near the same ship, seed points are selected by using a general method and a ship-type characteristic-based method respectively, and a streamline is generated. The following table compares the two methods in terms of seed point number, computation time, and streamline effect.
TABLE 1 comparison of the results of the experiments of the present invention and the general method
Figure SMS_91
。
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that changes may be made in the embodiments and/or equivalents thereof without departing from the spirit and scope of the invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (8)

1. A streamline seed point selecting method based on ship appearance characteristics is characterized by comprising the following steps:
step 1: collecting and constructing a sample data set;
step 1-1: selecting a ship model to obtain a longitudinal section of the ship model, selecting a first area boundary line, a second area boundary line, a third area boundary line and a fourth area boundary line, dividing an area between a ship longitudinal section profile line and the first boundary line into a first area, dividing an area between the first boundary line and the second boundary line into a second area, dividing an area between the second boundary line and the third boundary line into a third area, and dividing an area between the third boundary line and the fourth boundary line into a fourth area;
step 1-2: recording a longitudinal-middle section contour line as a boundary line 0, a first area boundary line as a boundary line 1, a second area boundary line as a boundary line 2, a third area boundary line as a boundary line 3 and a fourth area boundary line as a boundary line 4;
step 1-3: constructing a first sample data set by taking the boundary line 0 as an input and the boundary line 1 as an output; constructing a second sample data set by taking the boundary line 1 as input, the boundary line 2 as output, the boundary line 2 as input and the boundary line 3 as output; constructing a fourth sample data set by taking the boundary line 3 as an input and the boundary line 4 as an output; and divided into a test set and a training set;
step 2: constructing a BP neural network, and predicting all points on a boundary line of a next layer from all points of a profile line of a longitudinal-medial section or all points on a boundary line of a region; the BP neural network hidden layer has 3 layers, the number of input and output neurons is m, and m is equal to the number of points taken on the boundary line;
and step 3: respectively inputting the constructed test set data into a BP neural network for training, and constructing a first area boundary line network model, a second area boundary line network model, a third area boundary line network model and a fourth area boundary line network model;
and 4, step 4: and respectively inputting the test set data into the trained boundary line network model to obtain corresponding boundary lines.
2. The streamline seed point selection method based on ship appearance characteristics according to claim 1, wherein the first region boundary line is selected by the following steps:
acquiring streamline seed points at set intervals on a profile line of a longitudinal and middle section of a ship according to the total length of the ship, wherein the total length of the ship is the maximum horizontal distance between the head end and the tail end of the surface of the ship, which is parallel to a designed waterline;
calculating the minimum included angle between a connecting line between two adjacent points and a horizontal line of each point on the profile line of the longitudinal-middle section, and calculating the slope of a vertical line of the connecting line between two adjacent points;
according to the slope of the vertical line obtained by the connecting line of the two adjacent points of the point, the slope extends a distance from the point to the direction far away from the profile contour line of the ship; connecting all end points of the extended vertical line to form a first area boundary line; and the area between the first area boundary line and the longitudinal and middle section contour lines is a first area.
3. The streamline seed point selection method based on ship appearance characteristics according to claim 2,
the second area boundary line, the third area boundary line and the fourth area boundary line are selected as follows:
acquiring streamline seed points at a set interval on the boundary line of the previous area according to the maximum horizontal distance between the head end and the tail end of the previous area; said spacing is
Figure QLYQS_1
Wherein is present>
Figure QLYQS_2
Represents the overall length of the ship;
calculating the minimum included angle between a connecting line between two adjacent points of each point on the boundary line of the previous area and a horizontal line, and calculating the slope of a perpendicular line of the connecting line between two adjacent points;
according to the slope of a vertical line obtained by connecting two adjacent points of the point, extending a distance from the point to the direction far away from the contour line of the middle section of the ship by using the slope; connecting all the end points of the extended vertical line to form a boundary line of the current area; and the area between the current area boundary line and the previous area boundary line is the current area.
4. The streamline seed point selection method based on ship appearance characteristics according to claim 3, wherein the vertical line is extended as follows:
if the minimum included angle between the connecting line between two adjacent points of the point and the horizontal line is positioned in the interval of [0 degrees and 30 degrees ], the perpendicular line connecting the adjacent points passing through the point extends outwards for a distance of t;
if the minimum included angle between the connecting line between two adjacent points of the point and the horizontal line is within the interval of (30 degrees and 60 degrees), the perpendicular line between the connecting lines of the adjacent points is extended outwards by the distance of 1.25 x t through the point;
if the minimum included angle between the connecting line between two adjacent points of the point and the horizontal line is within the interval of (60 degrees and 90 degrees), the perpendicular line between the connecting lines of the adjacent points is extended outwards by the distance of 1.5 x t through the point;
when each point of the first area boundary line is selected, T =0.05 × T, when the second area boundary line is selected, T =0.1 × T, when the third area boundary line is selected, T =0.15 × T, and when the fourth area boundary line is selected, T =0.2 × T, wherein T represents the draught depth of the ship.
5. The streamline seed point selection method based on ship appearance characteristics according to claim 1, wherein the specific steps in step 3 are as follows:
step 3-1: initializing BP neural network parameters, and setting learning rate of BP neural network as
Figure QLYQS_4
，/>
Figure QLYQS_9
Is->
Figure QLYQS_12
The weight matrix of the BP neural network is compared with>
Figure QLYQS_6
And a bias matrix pick>
Figure QLYQS_8
Is set to->
Figure QLYQS_11
In between, a random matrix, is selected>
Figure QLYQS_14
Is the BP neural network->
Figure QLYQS_3
The fifth on the hidden layer>
Figure QLYQS_7
Weight on each neuron->
Figure QLYQS_10
Is the number ^ greater or lesser of the bp neural network>
Figure QLYQS_13
The fifth on the hidden layer>
Figure QLYQS_5
A bias value on an individual neuron;
step 3-2: inputting the constructed training set into a BP neural network, iterating P times of sample data in the BP neural network, and obtaining the minimum value of the error of the MSE function as
Figure QLYQS_17
Will >>
Figure QLYQS_18
And a target value->
Figure QLYQS_22
By contrast, if->
Figure QLYQS_16
Less than the target value>
Figure QLYQS_20
Then the iteration is stopped, if->
Figure QLYQS_21
Greater than the target value>
Figure QLYQS_23
Then the iteration continues until pick>
Figure QLYQS_15
Less than the target value>
Figure QLYQS_19
。
6. The method for selecting streamline seed points based on ship appearance characteristics according to claim 5, wherein the weight matrix
Figure QLYQS_24
；
Figure QLYQS_25
Is a weight matrix->
Figure QLYQS_26
Is/are>
Figure QLYQS_27
Sub-status value>
Figure QLYQS_28
Is->
Figure QLYQS_29
The first derivative of (a);
bias matrix
Figure QLYQS_30
；
In the formula (I), the compound is shown in the specification,
Figure QLYQS_31
is a bias matrix->
Figure QLYQS_32
Is/are>
Figure QLYQS_33
A sub-status value, <' > or>
Figure QLYQS_34
Is->
Figure QLYQS_35
The first derivative of (a).
7. The streamline seed point selection method based on ship appearance characteristics according to claim 5, wherein a sigmoid function is adopted as an activation function of a BP neural network, and a MSE function is adopted as a loss function mean square error function;
the sigmoid function is:
Figure QLYQS_36
；
in the formula (I), the compound is shown in the specification,
Figure QLYQS_37
represents the value of the neuron after calculation by the activation function, is/are>
Figure QLYQS_38
A jth input value representing an ith hidden layer;
the MSE function is:
Figure QLYQS_39
；
in the formula (I), the compound is shown in the specification,mis the number of points on the boundary line,
Figure QLYQS_40
is the coordinate of the ith point of the set of points on the current solution boundary line c, and>
Figure QLYQS_41
the coordinates of the ith point in the set of points on the current solved boundary line c are shown, the value of c is 1,2,3 and 4,
Figure QLYQS_42
is the variance value of the two point coordinate values, i.e.>
Figure QLYQS_43
The value of (c).
8. The streamline seed point selection method based on ship appearance characteristics according to claim 6,
Figure QLYQS_44
；
in the formula (I), the compound is shown in the specification,
Figure QLYQS_45
representing the total length of the ship, and c is the boundary line of the current solution. />
CN202310165021.8A 2023-02-27 2023-02-27 Streamline seed point selection method based on ship appearance characteristics Active CN115859485B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310165021.8A CN115859485B (en) 2023-02-27 2023-02-27 Streamline seed point selection method based on ship appearance characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310165021.8A CN115859485B (en) 2023-02-27 2023-02-27 Streamline seed point selection method based on ship appearance characteristics

Publications (2)

Publication Number Publication Date
CN115859485A true CN115859485A (en) 2023-03-28
CN115859485B CN115859485B (en) 2023-05-23

Family

ID=85658850

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310165021.8A Active CN115859485B (en) 2023-02-27 2023-02-27 Streamline seed point selection method based on ship appearance characteristics

Country Status (1)

Country Link
CN (1) CN115859485B (en)

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090150097A1 (en) * 2007-12-07 2009-06-11 Landmark Graphics Corporation, A Halliburton Company Systems and Methods For Utilizing Cell Based Flow Simulation Results to Calculate Streamline Trajectories
WO2014049346A1 (en) * 2012-09-26 2014-04-03 Bergen Teknologioverføring As Image filtering
CN109559369A (en) * 2018-12-04 2019-04-02 中国科学院地理科学与资源研究所 It is a kind of towards the ocean current vector field visualization method based on angleonly tracking for flowing to feature
US20200160061A1 (en) * 2017-12-11 2020-05-21 Zhuhai Da Hengqin Technology Development Co., Ltd. Automatic ship tracking method and system based on deep learning network and mean shift
CN111832660A (en) * 2020-07-23 2020-10-27 南京联云智能系统有限公司 Multi-target real-time detection method for mobile ship
US20210117796A1 (en) * 2020-12-01 2021-04-22 Harbin Engineering University Ship Motion Prediction Method Based on Long Short-Term Memory Network and Gaussian Process Regression
CN113159117A (en) * 2021-03-11 2021-07-23 中国科学院计算机网络信息中心 Streamline generation method and device
CN114936413A (en) * 2022-04-21 2022-08-23 哈尔滨工程大学 Ship body shape optimization neural network modeling method and ship body shape optimization method
CN115511752A (en) * 2022-10-31 2022-12-23 合肥埃科光电科技股份有限公司 BP neural network-based point coordinate distortion removal method and storage medium
US20220414891A1 (en) * 2019-09-09 2022-12-29 South China University Of Technology Method for automatic segmentation of fuzzy boundary image based on active contour and deep learning

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090150097A1 (en) * 2007-12-07 2009-06-11 Landmark Graphics Corporation, A Halliburton Company Systems and Methods For Utilizing Cell Based Flow Simulation Results to Calculate Streamline Trajectories
WO2014049346A1 (en) * 2012-09-26 2014-04-03 Bergen Teknologioverføring As Image filtering
US20200160061A1 (en) * 2017-12-11 2020-05-21 Zhuhai Da Hengqin Technology Development Co., Ltd. Automatic ship tracking method and system based on deep learning network and mean shift
CN109559369A (en) * 2018-12-04 2019-04-02 中国科学院地理科学与资源研究所 It is a kind of towards the ocean current vector field visualization method based on angleonly tracking for flowing to feature
US20220414891A1 (en) * 2019-09-09 2022-12-29 South China University Of Technology Method for automatic segmentation of fuzzy boundary image based on active contour and deep learning
CN111832660A (en) * 2020-07-23 2020-10-27 南京联云智能系统有限公司 Multi-target real-time detection method for mobile ship
US20210117796A1 (en) * 2020-12-01 2021-04-22 Harbin Engineering University Ship Motion Prediction Method Based on Long Short-Term Memory Network and Gaussian Process Regression
CN113159117A (en) * 2021-03-11 2021-07-23 中国科学院计算机网络信息中心 Streamline generation method and device
CN114936413A (en) * 2022-04-21 2022-08-23 哈尔滨工程大学 Ship body shape optimization neural network modeling method and ship body shape optimization method
CN115511752A (en) * 2022-10-31 2022-12-23 合肥埃科光电科技股份有限公司 BP neural network-based point coordinate distortion removal method and storage medium

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
QIN,XJ等: "Streamline Uniform Placement Algorithm With Dynamic Seed Points", 《IEEE ACCESS》 *
吕天耀: "基于流线相似性的流场并行可视化方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *
熊光正;黄智濒;戴志涛;杨武兵;: "一种数据驱动的三维流场流线特征化筛选方法", 北京邮电大学学报 *
苏萁;王逸斌;赵宁;: "基于机器学习算法的舰面流场预警系统研究", 测控技术 *
高明正;张火明;金尚忠;: "BP神经网络在船舶与海洋工程中的应用研究", 舰船科学技术 *

Also Published As

Publication number Publication date
CN115859485B (en) 2023-05-23

Similar Documents

Publication Publication Date Title
CN109993280B (en) Underwater sound source positioning method based on deep learning
US11452286B2 (en) Method of predicting central fishing ground of flying squid family ommastrephidae
CN110118639B (en) Three-dimensional digital ship model virtual resistance experiment method and system
CN109858523B (en) Shallow sea sound velocity profile inversion method based on neural network and ray theory
CN106022471A (en) Wavelet neural network model ship rolling real-time prediction method based on particle swarm optimization algorithm
CN110516318B (en) Airfoil design method based on radial basis function neural network proxy model
CN112488208B (en) Method for acquiring remaining life of island pillar insulator
CN113777931B (en) Icing wing type pneumatic model construction method, device, equipment and medium
CN109919082A (en) Modal identification method based on LSTM and EMD
CN111105136A (en) Grey level evaluation method based on normalized sample
CN113743577A (en) Fine grid data partition construction method and system for mesoscale vortex identification
CN109460874A (en) A kind of ariyoshi wave height prediction technique based on deep learning
CN113468466B (en) One-dimensional wave equation solving method based on neural network
CN111914487B (en) Data-free regional hydrological parameter calibration method based on antagonistic neural network
CN115859485A (en) Streamline seed point selection method based on ship appearance characteristics
WO2022032874A1 (en) Adversarial neural network-based hydrological parameter calibration method for data region
CN113722980A (en) Ocean wave height prediction method, system, computer equipment, storage medium and terminal
CN111626341B (en) Feature level information fusion method for underwater target identification
CN110782078A (en) Learning method for predicting mud discharging rate of trailing suction hopper dredger
CN112464367B (en) Appearance and structure two-layer design optimization method of autonomous underwater vehicle
JPH0949884A (en) Predicting method for local weather using neural network
Obayashi Pareto solutions of multipoint design of supersonic wings using evolutionary algorithms
CN111766627B (en) Self-adaptive smooth surface wave imaging method based on model resolution
CN108805981A (en) A kind of branch channel type delta front training image method for building up
CN115114842A (en) Rainstorm waterlogging event prediction method based on small sample transfer learning algorithm

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