CN111169016A - 3+2 shaft unsupported 3D printing manufacturing method for blade parts - Google Patents

3+2 shaft unsupported 3D printing manufacturing method for blade parts Download PDF

Info

Publication number
CN111169016A
CN111169016A CN201911310642.0A CN201911310642A CN111169016A CN 111169016 A CN111169016 A CN 111169016A CN 201911310642 A CN201911310642 A CN 201911310642A CN 111169016 A CN111169016 A CN 111169016A
Authority
CN
China
Prior art keywords
sub
printing
clustering
blade
initial clustering
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
CN201911310642.0A
Other languages
Chinese (zh)
Other versions
CN111169016B (en
Inventor
吴宝海
李成林
张莹
郑海
张阳
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.)
Northwestern Polytechnical University
Original Assignee
Northwestern Polytechnical University
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 Northwestern Polytechnical University filed Critical Northwestern Polytechnical University
Priority to CN201911310642.0A priority Critical patent/CN111169016B/en
Publication of CN111169016A publication Critical patent/CN111169016A/en
Application granted granted Critical
Publication of CN111169016B publication Critical patent/CN111169016B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
    • B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
    • B29C64/00—Additive manufacturing, i.e. manufacturing of three-dimensional [3D] objects by additive deposition, additive agglomeration or additive layering, e.g. by 3D printing, stereolithography or selective laser sintering
    • B29C64/30—Auxiliary operations or equipment
    • B29C64/386—Data acquisition or data processing for additive manufacturing
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B33—ADDITIVE MANUFACTURING TECHNOLOGY
    • B33Y—ADDITIVE MANUFACTURING, i.e. MANUFACTURING OF THREE-DIMENSIONAL [3-D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVE AGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3-D PRINTING, STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING
    • B33Y50/00—Data acquisition or data processing for additive manufacturing
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
    • B29L—INDEXING SCHEME ASSOCIATED WITH SUBCLASS B29C, RELATING TO PARTICULAR ARTICLES
    • B29L2031/00—Other particular articles
    • B29L2031/08—Blades for rotors, stators, fans, turbines or the like, e.g. screw propellers
    • B29L2031/082—Blades, e.g. for helicopters

Abstract

The invention discloses a 3+ 2-axis unsupported 3D printing manufacturing method for blade parts, which is used for solving the technical problem that the existing multi-axis 3D printing manufacturing method for the blade parts is complex. According to the technical scheme, a blade part is divided into a plurality of sub-blocks through an improved spectral clustering algorithm, compared with a Z plane simple direct division method adopted in the background technology, the similarity and the adjacent degree of blades in different Z directions are considered, so that the surface normal vectors of the divided sub-blocks are more uniform, the step effect during printing is avoided, the clustering number can be automatically determined according to the clustering algorithm, the defects of the traditional spectral clustering algorithm are overcome, meanwhile, an iterative algorithm based on two criteria can also automatically give out the optimal clustering number, the defect that the traditional spectral clustering algorithm needs to manually give out the clustering number is overcome, the sub-blocks are printed in the main printing direction calculated by the method, a supporting structure is avoided in the whole 3D printing and forming process of the blade, and the method is simple and reliable.

Description

3+2 shaft unsupported 3D printing manufacturing method for blade parts
Technical Field
The invention relates to a multi-axis 3D printing manufacturing method of blade parts, in particular to a 3+2 axis unsupported 3D printing manufacturing method of blade parts.
Background
At present, many researchers put emerging manufacturing technologies into the manufacturing research of blisks and impeller parts for obtaining manufacturing and processing methods with higher efficiency and lower cost, and 3D printing of the blisks and the impeller parts is one of effective solutions, so that the blisks and the impeller parts have the advantages of saving materials and time; on the other hand, the manufacturing of the blades is the key for printing and manufacturing the impeller parts of the blisk, but the blades belong to curved surface parts, so that the problems of complex structure, serious surface bending and twisting and the like exist, and when a 3-axis 3D printer is used for printing, the following two problems mainly exist; one is the sharp height error caused by the non-perpendicularity of the printing direction and the normal vector of the surface of the part, namely the 'step effect': the printed part surface has sharp height (see figure 3), and the ladder effect is more obvious when the non-perpendicularity degree is larger; secondly, the part suspension area may need to be added with a support to form the part, the introduction of the support structure will increase the time of the whole forming process, and in addition, the processing time is needed to remove the support structure, and meanwhile, the existence of the support structure also causes the waste of materials, which is obviously contrary to the original purpose of 3D printing; in addition, the process of removing the support structure can also result in a loss of surface quality at the support structure and part contact.
Document 1, "w.wang, c.zanni, and l.kobbelt," Improved surface quality In 3 printing by optimizing the printing direction, "In Computer Graphics Forum, vol.35, No.2.wiley Online Library, pp.59-70,2016," proposes to achieve low support and surface perpendicular printing by dividing parts for successive printing and finally manually butting "stitching", but is applied to thin-wall cranked heavy blade-like parts corresponding to the present invention, which have an unsatisfactory dividing effect and still require partial support.
Document 2, "chinese patent application publication No. CN 105149582A" discloses a multi-axis 3D printing and forming method for blade parts, but the blade is divided in a plane only in the Z direction, which is too simple to be suitable for the case of large bending degree, and the printing process still needs a certain support.
Disclosure of Invention
In order to overcome the defect that the existing multi-axis 3D printing manufacturing method of the blade parts is complex, the invention provides a 3+ 2-axis unsupported 3D printing manufacturing method of the blade parts. According to the method, the blade parts are divided into a plurality of sub-blocks through an improved spectral clustering algorithm, compared with a Z plane simple direct division method adopted in the background technology, the similarity and the adjacency degree of the blades in different Z directions are considered, so that the surface normal vectors of the divided sub-blocks are more uniform, the step effect during printing is avoided, the clustering number can be automatically determined according to the clustering algorithm, the defects of the traditional spectral clustering algorithm are overcome, meanwhile, an iterative algorithm based on two criteria can also automatically give out the optimal clustering number, the defect that the traditional spectral clustering algorithm needs to manually give out the clustering number is overcome, the sub-blocks are printed in the main printing direction calculated by the method, a supporting structure is avoided in the whole 3D printing and forming process of the blade, and the method is simple and reliable.
The technical scheme adopted by the invention for solving the technical problems is as follows: a3 +2 shaft unsupported 3D printing manufacturing method of blade parts is characterized by comprising the following steps:
step one, importing a leaf part three-dimensional model in an STL format, and simultaneously respectively importing 134 and 994 point sets which are distributed on a unit sphere in a grid mode to serve as a printing voting vector set D of an algorithm0And a set of candidate printing direction vectors D1；
Step two, performing over-segmentation on the triangular patch to realize the pretreatment of the model; according to the three-dimensional model of the blade to be manufactured, software measures to obtain the maximum thickness d of the blade, and the maximum collapse angle theta of the blade is found according to the printing material used by the partcThe included angle threshold value is used as whether the normal vector of the triangular patch votes for the voting printing direction vector or not;
traversing the whole set of triangular patches and the set of candidate printing direction vectors D1Calculating the included angle between the two, if a certain triangular patch TriAnd a certain candidate printing direction vector
Figure BDA0002324425060000021
Angle of inclusion at threshold thetacInternal, then TriTo pair
Figure BDA0002324425060000022
And (5) casting 1 ticket. After the voting process is finished, pair D1Classifying triangular patches with consistent printing direction vector conditions of medium votes together, then calculating whether the centroid distance between the classified triangular patches is less than or equal to 2D, if so, dividing the classified triangular patches into 1 initial clustering unit, otherwise, forming an initial clustering unit by the classified triangular patches respectively, and otherwise, if a certain triangular patch and other patch pairs D are in the same pair1If the vectors in the three-dimensional space vector have no same vote, the triangular patch alone forms 1 initial clustering unit;
thirdly, constructing a similarity matrix W for the given blade part B to describe the similarity relation between each initial clustering unit as a basis for next clustering segmentation, and then calculating a normalization matrix D to further obtain a normalized Laplacian matrix L of the blade part;
the size of the similarity matrix W is n2×n2Wherein n is2Element w as the number of initial clustering unitsijRepresents an initial clustering unit CiAnd CjA rule relation is established between the initial clustering units, and w is used for eliminating the self similarity of the initial clustering unit Cii＝0；
The similarity matrix establishment criterion of the blade parts is as follows:
step error is minimum: from the above, in the absence of a support structure, the error of the 3D printed part is caused by a step error caused by the fact that the printing direction and the normal vector of the surface of the part are not completely perpendicular to each other
Figure BDA0002324425060000023
Denotes the printing direction, h denotes the cross-sectional layer thickness, and n is the surface normal vector of the current printed part face. Definition of tip height ciComprises the following steps:
Figure BDA0002324425060000031
step error E of single initial clustering unitiComprises the following steps:
Ei＝ci·S(Ci) (2)
wherein S (C)i) Represents an initial clustering unit CiThe area of (d);
calculating an initial clustering unit CiAnd CjThe step error between is expressed as:
Figure BDA0002324425060000032
wherein the content of the first and second substances,
Figure BDA0002324425060000033
to minimize the step error, there are
Figure BDA0002324425060000034
Wherein the sum of the areas of the two is divided by S (C)i)+S(Cj) Homogenizing the area of the initial clustering unit;
the adjacent closeness degree between the adjacent initial clustering units is maximized: obviously, if two initial clustering units are to be divided into the same sub-block, the two initial clustering units must be adjacent, and under the condition that the step errors are the same, the greater the adjacent degree is, the more the priority is for dividing into the same sub-block, and any two adjacent initial clustering units C are divided intoi、CjDegree of adjacency Ne ofijIs represented as follows:
Figure BDA0002324425060000035
Figure BDA0002324425060000036
Figure BDA0002324425060000037
Figure BDA0002324425060000038
wherein x isiIs represented by CiSet of x coordinates of vertices, y, of all triangular patches in the seti、zi、xj、yj、zjFor the same reason, xijThen is Ci、CjThe x coordinate set of the vertexes of all the triangular patches;
through the two steps, the similarity matrix is established in the following way:
Figure BDA0002324425060000039
wherein wijIs the ith row and j column element, lambda in the similarity matrix1、λ2Respectively debugging parameters
The normalized matrix D is calculated as follows:
Figure BDA0002324425060000041
the normalized Laplace matrix L of the blade parts has the formula:
Figure BDA0002324425060000042
step four, clustering the initial clustering units of the similarity matrix by using an improved clustering algorithm to obtain the optimal sub-block division B (B)1,B2,…BN) Calculating a main printing direction vector of each sub-block and each parameter representing theoretical printing error;
converting the clustering problem between initial clustering units in the solved similarity matrix into N classification problems of the feature vectors f corresponding to the first N feature values with the minimum L by adopting an Ncut method;
then standardizing the characteristic matrix composed of the N characteristic vectors into F according to rows, wherein each row is used as a N-dimensional sample, m samples are totally obtained, K-means clustering is adopted, the clustering dimension is N, and finally the optimal subblock division B (B) is obtained1,B2,…BN)；
Calculating a main printing direction vector of each sub-block: firstly, traversing 994 candidate printing direction vector sets D generated before for each sub-block in turn0Taking out one element of the
Figure BDA0002324425060000043
Satisfy the requirement of
Figure BDA0002324425060000044
The error of the included angle value of all triangular patches in the corresponding sub-block is
Figure BDA0002324425060000045
The minimum error value of the included angle between the sub-block and the sub-block;
various parameters representing theoretical printing errors are calculated: the method specifically comprises the steps that the sum of included angle values of normal vectors of each triangular surface patch in all sub-blocks and main printing direction vectors of the corresponding sub-blocks is recorded as sum _ ang, the average value of the total included angle, ave _ ang and the variance of the total included angle, var _ ang and the maximum value of the included angle in a part is recorded as ext _ ang, and the expression is as follows;
Figure BDA0002324425060000046
Figure BDA0002324425060000047
Figure BDA0002324425060000048
Figure BDA0002324425060000049
ext_ang＝max(angij) (16)
for convenient postamble expression, unify these parameters to the parameter set Par of the output result;
step five, determining the number of the optimal clustering sub-blocks, the theoretical printing error and the blade segmentation information under the current clustering number by an iterative algorithm;
two criteria for the iterative algorithm to perform:
the included angle between the normal vector of all triangular patches in the sub-block and the main printing direction of the sub-block is less than the material collapse angle thetac；
The formula for calculating the included angle is as follows:
Figure BDA0002324425060000051
wherein, thetaiExpressing the normal vector N of any triangular patch when the number of clusters is NiCorresponding to the main printing direction PiThe included angle of (A);
theoretical printing error E of each initial clustering unittAll within a given manufacturing tolerance requirement E of the bladedWithin the range;
if the blade is divided into 2 sub-blocks B1、B2Then, the algorithm divides the initial clustering units included in each sub-block, and B is assumed1In m initial clustering units, B2N, while giving the main printing direction vector P of each sub-blocki(ii) a Calculating theoretical printing error of any one initial clustering unit of the blades by the following formula
Figure BDA0002324425060000052
At this time j ∈ [1, m + n ]]
Figure BDA0002324425060000053
The specific flow of the iterative algorithm is as follows:
the clustering number N is an initial value, the clustering number N is substituted into a clustering algorithm to obtain output of a partitioning result of an initial clustering unit, an output result parameter set Par and the like, if elements in the current Par are all smaller than the Par output by the last N value, the next step is carried out, otherwise, clustering partitioning is carried out again until the condition is met;
two iteration criteria are judged:
Figure BDA0002324425060000054
θi≤θcif yes, outputting a clustering result, wherein the clustering number at the moment is the optimal clustering number N; if not, N is equal to N +1, and then the step four is returned to cluster again until the condition is met;
and step six, applying the result output by the algorithm to 3+ 2-axis 3D printing of the blade part to complete the solid printing of the blade part.
When each subblock is printed, the printer needs to adjust the position and posture of the printed part, so that the main printing direction of the current printed subblock given by the algorithm is consistent with the printing direction of the multi-axis 3D printer, the subblocks divided by the clustering algorithm are sequentially printed from bottom to top, and the unsupported high-quality blocked 3D multi-axis printing of the blade parts is realized.
The invention has the beneficial effects that: according to the method, the blade parts are divided into a plurality of sub-blocks through an improved spectral clustering algorithm, compared with a Z plane simple direct division method adopted in the background technology, the similarity and the adjacency degree of the blades in different Z directions are considered, so that the surface normal vectors of the divided sub-blocks are more uniform, the step effect during printing is avoided, the clustering number can be automatically determined according to the clustering algorithm, the defects of the traditional spectral clustering algorithm are overcome, meanwhile, an iterative algorithm based on two criteria can also automatically give out the optimal clustering number, the defect that the traditional spectral clustering algorithm needs to manually give out the clustering number is overcome, the sub-blocks are printed in the main printing direction calculated by the method, a supporting structure is avoided in the whole 3D printing and forming process of the blade, and the method is simple and reliable.
Specifically, the invention combines the structural characteristics of the blade parts and the technological characteristics of multi-axis 3D printing, improves the spectral clustering algorithm of the background technology, enables the spectral clustering algorithm to be applied to the STL model of the blade parts, and performs segmentation according with the 3D printing technology;
when the triangular patch of the STL model of the blade part is processed, the triangular patch is initially clustered into a smaller number of initial clustering units by adopting an over-segmentation step, so that the efficiency of a subsequent clustering algorithm is effectively improved;
the invention also provides a method for automatically giving the optimal clustering number according to the blade part additive manufacturing process, and overcomes the defect that the clustering number needs to be given by the traditional spectral clustering algorithm;
according to the invention, the method of sub-block step-by-step printing is adopted for 3D printing of the blades, and each sub-block is printed in the main printing direction according to the algorithm, so that the problems that a large number of supporting structures are needed in the printing process and the printing quality of partial areas is poor due to the fact that the blade parts are generally bent seriously are solved.
The method can automatically divide the blade into a plurality of sub-blocks for realizing optimal printing, and has the advantages of no support and high printing quality compared with the traditional 3-axis 3D printing.
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
Drawings
FIG. 1 is a flow chart of a 3+ 2-axis unsupported 3D printing manufacturing method of the blade type part.
FIG. 2 is a schematic representation of the results of the blade division obtained by the method of the present invention.
FIG. 3 is a schematic diagram of a step effect formed by a prior art method.
Detailed Description
Reference is made to fig. 1-2. The 3+2 shaft unsupported 3D printing manufacturing method of the blade parts comprises the following specific steps:
step 1, preparation: introducing a leaf part three-dimensional model in STL format (675 triangular patches in total), and respectively introducing 134 and 994 point sets distributed in a grid on a unit sphere as a printing voting vector set D of an algorithm0And a set of candidate printing direction vectors D1；
Step 2, segmentation pretreatment: because the calculated amount of the segmentation algorithm directly acted by the triangular patch of the blade model as the minimum unit is huge and redundant, the model is preprocessed firstly to improve the algorithm efficiency.
According to the three-dimensional model of the blade to be manufactured, software measures the maximum thickness d of the blade, wherein d is 18mm, and the maximum collapse angle | delta theta is found according to the printing material used by the partcIf the angle is 30 degrees, the angle is used as an included angle threshold value of whether the normal vector of the triangular patch votes for the voting printing direction vector;
traversing the whole set of triangular patches and the set of candidate printing direction vectors D1Calculating the included angle between the two, if a certain triangular patch TriAnd a certain candidate printing direction vector
Figure BDA0002324425060000079
Angle of inclusion at threshold thetacIn, i.e.
Figure BDA0002324425060000071
Figure BDA0002324425060000072
Then TriTo pair
Figure BDA0002324425060000073
And (5) casting 1 ticket. After the voting process is finished, pair D1Classifying the triangular patches with consistent printing direction vector conditions of the medium votes together, then calculating whether the centroid distance between the triangular patches is less than or equal to 36mm, if so, dividing the classified triangular patches into 1 initial clustering unit, otherwise, independently forming an initial clustering unit between the classified triangular patches, and otherwise, if a certain triangular patch and other patch pairs D are in charge of1If the vectors in the three-dimensional space vector have no same vote, the triangular patch alone forms 1 initial clustering unit, and finally 279 initial clustering units are obtained;
step 3, constructing a similarity matrix and a Laplace matrix of the triangular patch of the blade: for a given blade part B, constructing a similarity matrix W for describing the magnitude of similarity relation among initial clustering units as a basis for next clustering segmentation, and then calculating a normalization matrix D to further obtain a Laplacian matrix L for blade part normalization;
the size of the similarity matrix W is n2×n2(wherein n is2279, the number of initial clustering units), wherein the element wijRepresents an initial clustering unit CiAnd CjA criterion relation (kernel function) is established between the two, and w is used for eliminating the similarity of the initial clustering unit Cii＝0；
The similarity matrix establishment criterion of the blade parts is as follows:
step error is minimum: from the above, in the absence of a support structure, the error of a 3D printed part is caused by a step error (step effect) generated by incomplete perpendicularity of the printing direction and the normal vector of the surface of the part, wherein the step error is generated
Figure BDA0002324425060000074
The printing direction is shown, h represents the thickness of the cross-sectional layer, 0.3 is taken, and n is the surface normal vector of the current printed part surface.
Definition of tip height ciComprises the following steps:
Figure BDA0002324425060000075
step error E of single initial clustering unitiComprises the following steps:
Ei＝ci·S(Ci) (2)
wherein, S (C)i) Represents an initial clustering unit CiThe area of (d);
calculating an initial clustering unit CiAnd CjThe step error between is expressed as:
Figure BDA0002324425060000076
wherein the content of the first and second substances,
Figure BDA0002324425060000077
to minimize the step error, there are
Figure BDA0002324425060000078
Wherein, the sum of the areas of the two is divided by S (C)i)+S(Cj) The area of the initial clustering unit is homogenized;
the adjacent closeness degree between the adjacent initial clustering units is maximized: obviously, if two initial clustering units are to be divided into the same sub-block, the initial clustering units must be adjacent to each other, and under the condition that the step errors are the same, the greater the adjacent degree is, the more the priority is for dividing the two initial clustering units into the same sub-block, and any two adjacent initial clustering units C are divided intoi、CjDegree of adjacency Ne ofijIs represented as follows:
Figure BDA0002324425060000081
Figure BDA0002324425060000082
Figure BDA0002324425060000083
Figure BDA0002324425060000084
wherein x isiIs represented by CiSet of x coordinates of vertices, y, of all triangular patches in the seti、zi、xj、yj、zjFor the same reason, xijThen is Ci、CjThe x coordinate set of the vertexes of all the triangular patches;
through the two steps, the similarity matrix is established in the following way:
Figure BDA0002324425060000085
wherein λ is1Take 0.7, λ2Take 0.3
The normalized matrix D is calculated as follows:
Figure BDA0002324425060000086
the normalized Laplace matrix L of the blade parts has the formula:
Figure BDA0002324425060000087
and 4, dividing triangular patches of the blade parts: clustering the initial clustering units of the similarity matrix by using an improved clustering algorithm to obtain the optimal sub-block division B (B)1,B2,…BN) Calculating a main printing direction vector of each sub-block and each parameter representing theoretical printing error;
converting the clustering problem between initial clustering units in the solved similarity matrix into N classification problems of the feature vectors f corresponding to the first N feature values with the minimum L by adopting an Ncut method;
then standardizing the characteristic matrix composed of the N characteristic vectors into F according to rows, wherein each row is used as a N-dimensional sample, m samples are totally obtained, K-means clustering is adopted, the clustering dimension is N, and finally the optimal subblock division B (B) is obtained1,B2,…BN)；
Calculating a main printing direction vector of each sub-block: traversing the 994 candidate printing direction vector sets D generated before for each sub-block in turn0Taking out one element of the
Figure BDA0002324425060000091
Satisfy the requirement of
Figure BDA0002324425060000092
The error of the included angle value of all triangular patches in the corresponding sub-block is
Figure BDA0002324425060000093
The minimum error value of the included angle between the sub-block and the sub-block;
various parameters representing theoretical printing errors are calculated: specifically, the method comprises the following steps of obtaining information such as the sum (recorded as sum _ ang) of included angle values of normal vectors of each triangular patch in all sub-blocks and main printing direction vectors of the corresponding sub-blocks, the average (recorded as ave _ ang) of total included angles, the variance (recorded as var _ ang) of the total included angles, the maximum (recorded as ext _ ang) of included angles in parts and the like, wherein the expression is as follows;
Figure BDA0002324425060000094
Figure BDA0002324425060000095
Figure BDA0002324425060000096
Figure BDA0002324425060000097
ext_ang＝max(angij) (16)
for convenient expression, unifying the parameters into an output result parameter set Par;
step 5, determining the number of the optimal clustering sub-blocks, theoretical printing errors and blade segmentation information under the current clustering number by an iterative algorithm;
two criteria for the iterative algorithm to perform:
the normal vectors of all triangular patches in a sub-block are all less than the material collapse angle thetac＝30°；
The formula for calculating the included angle is as follows:
Figure BDA0002324425060000098
wherein, thetaiExpressing the normal vector N of any triangular patch when the number of clusters is NiAnd correspond toMain printing direction PiThe included angle of (A);
to ensure the quality of the manufactured blade, a theoretical printing error E of each initial clustering unit is specifiedtAll within a given manufacturing tolerance requirement E of the bladed20 degrees;
if the blade is divided into 2 sub-blocks B1、B2The algorithm divides the initial clustering unit included in each sub-block (assume B)1In m initial clustering units, B2N) in the sub-block while giving a main printing direction vector P of each sub-blocki(ii) a Calculating theoretical printing error of any one initial clustering unit of the blades by the following formula
Figure BDA0002324425060000101
At this time j ∈ [1, m + n ]]
Figure BDA0002324425060000102
The specific flow of the iterative algorithm is as follows:
the clustering number N is an initial value, the clustering number N is substituted into a clustering algorithm to obtain output of a partitioning result of an initial clustering unit, an output result parameter set Par and the like, if elements in the current Par are all smaller than the Par output by the last N value, the next step is carried out, otherwise, clustering partitioning is carried out again until the condition is met;
two iteration criteria are judged:
Figure BDA0002324425060000103
θi≤θcif yes, outputting a clustering result, wherein the clustering number at the moment is the optimal clustering number N; if not, N is equal to N +1, and then the step 4 is returned to cluster again until the condition is met;
when the iteration is carried out until the N is 3, the conditions are met, namely, the impeller blade is divided into 3 sub-blocks, and sum _ ang is 4092.4622 degrees, the average value ave _ ang is 6.0943 degrees, the variance var _ ang is 20.1656 degrees, the maximum value ext _ ang in the part is 19.935 degrees, and the main printing directions of the 4 sub-blocks are P1(-0.1379,0.6935,0.7071)、P2(0，-0.9569，-0.2903)、P3(-0.1379，0.6935，0.7071)、P4(0.1622，0.8155，0.5556)
Step 6, applying the result output by the algorithm to 3+ 2-axis 3D printing of the blade part to complete the solid printing of the blade part;
when each subblock is printed, the 3+ 2-axis 3D printer needs to adjust the position and posture of the printed part, so that the main printing direction of the currently printed subblock given by the algorithm is consistent with the printing direction of the multi-axis 3D printer, the subblocks divided by the clustering algorithm are sequentially printed from bottom to top, and the unsupported high-quality blocked 3D multi-axis printing of the blade parts is realized.

Claims (1)

1. A3 +2 shaft unsupported 3D printing manufacturing method of a blade part is characterized by comprising the following steps:
step one, importing a leaf part three-dimensional model in an STL format, and simultaneously respectively importing 134 and 994 point sets which are distributed on a unit sphere in a grid mode to serve as a printing voting vector set D of an algorithm0And a set of candidate printing direction vectors D1；
Step two, performing over-segmentation on the triangular patch to realize the pretreatment of the model; according to the three-dimensional model of the blade to be manufactured, software measures to obtain the maximum thickness d of the blade, and the maximum collapse angle theta of the blade is found according to the printing material used by the partcThe included angle threshold value is used as whether the normal vector of the triangular patch votes for the voting printing direction vector or not;
traversing the whole set of triangular patches and the set of candidate printing direction vectors D1Calculating the included angle between the two, if a certain triangular patch TriAnd a certain candidate printing direction vector
Figure FDA0002324425050000011
Angle of inclusion at threshold thetacInternal, then TriTo pair
Figure FDA0002324425050000012
1, casting a ticket; throw-inAfter the ticket process is finished, the pair D is processed1Classifying triangular patches with consistent printing direction vector conditions of medium votes together, then calculating whether the centroid distance between the classified triangular patches is less than or equal to 2D, if so, dividing the classified triangular patches into 1 initial clustering unit, otherwise, forming an initial clustering unit by the classified triangular patches respectively, and otherwise, if a certain triangular patch and other patch pairs D are in the same pair1If the vectors in the three-dimensional space vector have no same vote, the triangular patch alone forms 1 initial clustering unit;
thirdly, constructing a similarity matrix W for the given blade part B to describe the similarity relation between each initial clustering unit as a basis for next clustering segmentation, and then calculating a normalization matrix D to further obtain a normalized Laplacian matrix L of the blade part;
the size of the similarity matrix W is n2×n2Wherein n is2Element w as the number of initial clustering unitsijRepresents an initial clustering unit CiAnd CjA rule relation is established between the initial clustering units, and w is used for eliminating the self similarity of the initial clustering unit Cii＝0；
The similarity matrix establishment criterion of the blade parts is as follows:
step error is minimum: from the above, in the absence of a support structure, the error of the 3D printed part is caused by a step error caused by the fact that the printing direction and the normal vector of the surface of the part are not completely perpendicular to each other
Figure FDA0002324425050000013
Representing the printing direction, h represents the thickness of the cross-sectional layer, and n is the surface normal vector of the current printing part surface; definition of tip height ciComprises the following steps:
Figure FDA0002324425050000014
step error E of single initial clustering unitiComprises the following steps:
Ei＝ci·S(Ci) (2)
wherein S (C)i) Represents an initial clustering unit CiThe area of (d);
calculating an initial clustering unit CiAnd CjThe step error between is expressed as:
Figure FDA0002324425050000021
wherein the content of the first and second substances,
Figure FDA0002324425050000022
to minimize the step error, there are
Figure FDA0002324425050000023
Wherein the sum of the areas of the two is divided by S (C)i)+S(Cj) Homogenizing the area of the initial clustering unit;
the adjacent closeness degree between the adjacent initial clustering units is maximized: obviously, if two initial clustering units are to be divided into the same sub-block, the two initial clustering units must be adjacent, and under the condition that the step errors are the same, the greater the adjacent degree is, the more the priority is for dividing into the same sub-block, and any two adjacent initial clustering units C are divided intoi、CjDegree of adjacency Ne ofijIs represented as follows:
Figure FDA0002324425050000024
Figure FDA0002324425050000025
Figure FDA0002324425050000026
Figure FDA0002324425050000027
wherein x isiIs represented by CiSet of x coordinates of vertices, y, of all triangular patches in the seti、zi、xj、yj、zjFor the same reason, xijThen is Ci、CjThe x coordinate set of the vertexes of all the triangular patches;
through the two steps, the similarity matrix is established in the following way:
Figure FDA0002324425050000028
wherein wijIs the ith row and j column element, lambda in the similarity matrix1、λ2Respectively debugging parameters
The normalized matrix D is calculated as follows:
Figure FDA0002324425050000029
the normalized Laplace matrix L of the blade parts has the formula:
Figure FDA00023244250500000210
step four, clustering the initial clustering units of the similarity matrix by using an improved clustering algorithm to obtain the optimal sub-block division B (B)1,B2,…BN) Calculating a main printing direction vector of each sub-block and each parameter representing theoretical printing error;
converting the clustering problem between initial clustering units in the solved similarity matrix into N classification problems of the feature vectors f corresponding to the first N feature values with the minimum L by adopting an Ncut method;
then, standardizing the feature matrix composed of the N feature vectors into F according to rows, wherein each row is used as an N-dimensional sample, m samples are totally collected by adopting K-means clusteringClass dimension is N, and the optimal sub-block division B (B) is obtained1,B2,…BN)；
Calculating a main printing direction vector of each sub-block: firstly, traversing 994 candidate printing direction vector sets D generated before for each sub-block in turn0Taking out one element of the
Figure FDA0002324425050000031
Satisfy the requirement of
Figure FDA0002324425050000032
The error of the included angle value of all triangular patches in the corresponding sub-block is
Figure FDA0002324425050000033
The minimum error value of the included angle between the sub-block and the sub-block;
various parameters representing theoretical printing errors are calculated: the method specifically comprises the steps that the sum of included angle values of normal vectors of each triangular surface patch in all sub-blocks and main printing direction vectors of the corresponding sub-blocks is recorded as sum _ ang, the average value of the total included angle, ave _ ang and the variance of the total included angle, var _ ang and the maximum value of the included angle in a part is recorded as ext _ ang, and the expression is as follows;
Figure FDA0002324425050000034
Figure FDA0002324425050000035
Figure FDA0002324425050000036
Figure FDA0002324425050000037
ext_ang＝max(angij) (16)
for convenient postamble expression, unify these parameters to the parameter set Par of the output result;
step five, determining the number of the optimal clustering sub-blocks, the theoretical printing error and the blade segmentation information under the current clustering number by an iterative algorithm;
two criteria for the iterative algorithm to perform:
the included angle between the normal vector of all triangular patches in the sub-block and the main printing direction of the sub-block is less than the material collapse angle thetac；
The formula for calculating the included angle is as follows:
Figure FDA0002324425050000038
wherein, thetaiExpressing the normal vector N of any triangular patch when the number of clusters is NiCorresponding to the main printing direction PiThe included angle of (A);
theoretical printing error E of each initial clustering unittAll within a given manufacturing tolerance requirement E of the bladedWithin the range;
if the blade is divided into 2 sub-blocks B1、B2Then, the algorithm divides the initial clustering units included in each sub-block, and B is assumed1In m initial clustering units, B2N, while giving the main printing direction vector P of each sub-blocki(ii) a Calculating theoretical printing error of any one initial clustering unit of the blades by the following formula
Figure FDA0002324425050000041
At this time j ∈ [1, m + n ]]
Figure FDA0002324425050000042
The specific flow of the iterative algorithm is as follows:
the clustering number N is an initial value, the clustering number N is substituted into a clustering algorithm to obtain output of a partitioning result of an initial clustering unit, an output result parameter set Par and the like, if elements in the current Par are all smaller than the Par output by the last N value, the next step is carried out, otherwise, clustering partitioning is carried out again until the condition is met;
two iteration criteria are judged:
Figure FDA0002324425050000043
θi≤θcif yes, outputting a clustering result, wherein the clustering number at the moment is the optimal clustering number N; if not, N is equal to N +1, and then the step four is returned to cluster again until the condition is met;
step six, applying the result output by the algorithm to 3+ 2-axis 3D printing of the blade part to complete the solid printing of the blade part;
when each subblock is printed, the printer needs to adjust the position and posture of the printed part, so that the main printing direction of the current printed subblock given by the algorithm is consistent with the printing direction of the multi-axis 3D printer, the subblocks divided by the clustering algorithm are sequentially printed from bottom to top, and the unsupported high-quality blocked 3D multi-axis printing of the blade parts is realized.
CN201911310642.0A 2019-12-18 2019-12-18 3+2 shaft unsupported 3D printing manufacturing method for blade parts Active CN111169016B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911310642.0A CN111169016B (en) 2019-12-18 2019-12-18 3+2 shaft unsupported 3D printing manufacturing method for blade parts

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911310642.0A CN111169016B (en) 2019-12-18 2019-12-18 3+2 shaft unsupported 3D printing manufacturing method for blade parts

Publications (2)

Publication Number Publication Date
CN111169016A true CN111169016A (en) 2020-05-19
CN111169016B CN111169016B (en) 2021-11-16

Family

ID=70652048

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911310642.0A Active CN111169016B (en) 2019-12-18 2019-12-18 3+2 shaft unsupported 3D printing manufacturing method for blade parts

Country Status (1)

Country Link
CN (1) CN111169016B (en)

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1843485A1 (en) * 2006-03-30 2007-10-10 Sony Deutschland Gmbh Multiple-input multiple-output (MIMO) spatial multiplexing system with dynamic antenna beam combination selection capability
US20100142805A1 (en) * 2008-12-05 2010-06-10 Tandent Vision Science, Inc. Constraint generation for use in image segregation
CN103336969A (en) * 2013-05-31 2013-10-02 中国科学院自动化研究所 Image meaning parsing method based on soft glance learning
CN103400386A (en) * 2013-07-30 2013-11-20 清华大学深圳研究生院 Interactive image processing method used for video
CN103963303A (en) * 2014-04-30 2014-08-06 杭州先临三维科技股份有限公司 3D printing method and system
CN106650151A (en) * 2016-12-30 2017-05-10 湖南航天新材料技术研究院有限公司 Model for obtaining structure design parameters
CN106903889A (en) * 2015-12-21 2017-06-30 姜海涛 A kind of rapid section algorithm of STL models
US20170217103A1 (en) * 2016-01-29 2017-08-03 Massachusetts Institute Of Technology Techniques for color contoning in additive fabrication and related systems and methods
EP3232351A1 (en) * 2016-04-12 2017-10-18 Sergei F. Burlatsky System and process for evaluating and validating additive manufacturing operations
WO2018085507A1 (en) * 2016-11-04 2018-05-11 Cc3D Llc Additive manufacturing system having gravity-fed matrix
CN109466065A (en) * 2018-11-05 2019-03-15 山东大学 Exempt to support non-homogeneous three periods minimal surface generation method and system towards 3D printing
US10354304B2 (en) * 2013-03-15 2019-07-16 Kemeera Inc. Systems and methods for creating 3D objects
CN110450401A (en) * 2019-09-05 2019-11-15 上海交通大学医学院附属第九人民医院 A kind of method of biological 3 D-printing joint prosthesis in support liquid medium
CN110522501A (en) * 2019-08-09 2019-12-03 常州市第二人民医院 The building of 3D printing personalization orthopedic implant and biomethanics optimized treatment method

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1843485A1 (en) * 2006-03-30 2007-10-10 Sony Deutschland Gmbh Multiple-input multiple-output (MIMO) spatial multiplexing system with dynamic antenna beam combination selection capability
US20100142805A1 (en) * 2008-12-05 2010-06-10 Tandent Vision Science, Inc. Constraint generation for use in image segregation
US10354304B2 (en) * 2013-03-15 2019-07-16 Kemeera Inc. Systems and methods for creating 3D objects
CN103336969A (en) * 2013-05-31 2013-10-02 中国科学院自动化研究所 Image meaning parsing method based on soft glance learning
CN103400386A (en) * 2013-07-30 2013-11-20 清华大学深圳研究生院 Interactive image processing method used for video
CN103963303A (en) * 2014-04-30 2014-08-06 杭州先临三维科技股份有限公司 3D printing method and system
CN106903889A (en) * 2015-12-21 2017-06-30 姜海涛 A kind of rapid section algorithm of STL models
US20170217103A1 (en) * 2016-01-29 2017-08-03 Massachusetts Institute Of Technology Techniques for color contoning in additive fabrication and related systems and methods
EP3232351A1 (en) * 2016-04-12 2017-10-18 Sergei F. Burlatsky System and process for evaluating and validating additive manufacturing operations
WO2018085507A1 (en) * 2016-11-04 2018-05-11 Cc3D Llc Additive manufacturing system having gravity-fed matrix
CN106650151A (en) * 2016-12-30 2017-05-10 湖南航天新材料技术研究院有限公司 Model for obtaining structure design parameters
CN109466065A (en) * 2018-11-05 2019-03-15 山东大学 Exempt to support non-homogeneous three periods minimal surface generation method and system towards 3D printing
CN110522501A (en) * 2019-08-09 2019-12-03 常州市第二人民医院 The building of 3D printing personalization orthopedic implant and biomethanics optimized treatment method
CN110450401A (en) * 2019-09-05 2019-11-15 上海交通大学医学院附属第九人民医院 A kind of method of biological 3 D-printing joint prosthesis in support liquid medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
龚运息,陈晨,夏名祥,宋恩权: "FDM 3D打印模型表面阶梯效应的分析", 《制造技术与机床》 *

Also Published As

Publication number Publication date
CN111169016B (en) 2021-11-16

Similar Documents

Publication Publication Date Title
CN108312548B (en) Five-axis linkage 3D printing method based on model surface feature hybrid adaptive slicing
US11535400B2 (en) Fairing skin repair method based on measured wing data
CN111178432A (en) Weak supervision fine-grained image classification method of multi-branch neural network model
CN111709092B (en) Radial tire tread pattern finite element automatic modeling method based on finite element unit nodes
CN112053426B (en) Deep learning-based large-scale three-dimensional rivet point cloud extraction method
CN110349159B (en) Three-dimensional shape segmentation method and system based on weight energy adaptive distribution
CN104463871A (en) Streetscape facet extraction and optimization method based on vehicle-mounted LiDAR point cloud data
Zhou et al. Geometry and learning co-supported normal estimation for unstructured point cloud
US20210331815A1 (en) Method for controlling gap distribution of wing-fuselage joining based on measured data
CN113269791B (en) Point cloud segmentation method based on edge judgment and region growth
CN112364805B (en) Rotary palm image detection method
CN111169016B (en) 3+2 shaft unsupported 3D printing manufacturing method for blade parts
CN112164145A (en) Method for rapidly extracting indoor three-dimensional line segment structure based on point cloud data
CN112942837B (en) Cantilever structure concrete 3D printing method and system
CN109214269B (en) Human face posture alignment method based on manifold alignment and multi-image embedding
CN112270746A (en) Aluminum alloy 3D printing point cloud simplification algorithm based on neighborhood covariance characteristic parameter threshold
CN115797551B (en) Automatic modeling method for laser point cloud data based on two-step unsupervised clustering algorithm
CN116664889A (en) Aircraft flap skin repair allowance determination method
Mansouri et al. Segmentation-based semi-regular remeshing of 3D models using curvature-adapted subdivision surface fitting
CN114998103A (en) Point cloud cultural relic fragment three-dimensional virtual splicing method based on twin network
CN114004952A (en) Data processing method of point cloud with high-density noise based on statistical manifold curvature
CN114241026A (en) Point cloud simplification algorithm and device based on flatness division
CN114937011A (en) Photovoltaic cell image anomaly detection method based on empirical Garbor wavelet transform
CN112509018B (en) Quaternion space optimized three-dimensional image registration method
CN109410333B (en) High-quality super-patch clustering generation method

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