TWI777795B - A computer program product for evaluating sheet metal segmentation of three-dimensional sheet metal models - Google Patents

Abstract

A computer program product for sheet metal segmentation for evaluating three-dimensional sheet metal models, comprising a segmentation decision module, a flattening decision module and an evaluation module, wherein the segmentation decision module is found in the plate portion of the three-dimensional sheet metal model. Several initial plate parts are obtained, and the initial plate parts are respectively added to other adjacent plate parts to generate a plurality of three-dimensional sub-plate models. The flattening decision module flattens the three-dimensional sub-plate models to generate a plurality of flattening schemes. The evaluation module arranges the flattened sub-plate models in each of the flattening schemes on a plate body model; the evaluation module calculates a total corresponding to each of the flattening schemes according to each of the plate body models and a production cost parameter cost, and output a plate model with a lower total cost.

Description

A computer program product for evaluating sheet metal segmentation of three-dimensional sheet metal models

The invention relates to a computer program for sheet metal division; in particular, it refers to a computer program product for evaluating sheet metal division of a three-dimensional sheet metal model.

Sheet metal products are composed of multiple sheet metal parts. It is known that in the development process of sheet metal products, developers will create 3D sheet metal models of sheet metal products through 3D engineering software, so as to modify and optimize the structure and appearance of sheet metal products. etc., or make a virtual display of sheet metal products.

After the developers have confirmed the 3D sheet metal model, the next step is to remove the sheet metal. Since the stripping is the position where each sheet metal part of the sheet metal product is designated by the developer in the three-dimensional sheet metal model of the three-dimensional engineering software. That is, it takes a lot of time to divide the sheet metal in the three-dimensional sheet metal model based on the experience of the personnel, resulting in a situation of inefficiency.

Furthermore, the position of sheet metal division will affect the process cost of subsequent sheet metal parts processing. For inexperienced personnel, it is difficult to choose a better division position, resulting in an increase in processing costs.

In view of this, the purpose of the present invention is to provide a computer program product for evaluating the sheet metal division of the three-dimensional sheet metal model, which can increase the efficiency of evaluating the sheet metal division of the three-dimensional sheet metal model.

In order to achieve the above object, the present invention provides a computer program product for sheet metal segmentation for evaluating three-dimensional sheet metal models, comprising a segmentation decision module, a flattening decision module and an evaluation module, wherein the segmentation decision module The group is used to receive a three-dimensional drawing file, the three-dimensional drawing file includes a three-dimensional sheet metal model, and the three-dimensional sheet metal model includes a plurality of spliced plate parts; the segmentation decision module finds out the plate parts of the three-dimensional sheet metal model The plural ones are used as a plurality of initial plate parts, and other plate parts adjacent to each of the initial plate parts are respectively added from the initial plate parts to generate a plurality of three-dimensional sub-plate models, wherein the three-dimensional sub-plate models can be assembled Connecting into the three-dimensional sheet metal model; the flattening decision module receives the three-dimensional sub-sheet metal models, and flattens the three-dimensional sub-sheet metal models to generate a plurality of flattening schemes, each of which includes a plurality of flattening sub-sections Sheet metal models, each of the flattened sub-sheet models in each of the flattening schemes corresponds to each of the three-dimensional sub-sheet models, wherein at least a part of the three-dimensional sub-sheet models are flattened in different ways to form flat sheets in different flattening schemes. Flat sub-plate model; the evaluation module receives the flattening schemes, and arranges the flattened sub-plate models in each of the flattening schemes on a plate body model; the evaluation module is based on each of the plate body models and a production The cost parameter calculates a total cost corresponding to each of the flattening schemes, and outputs at least one of the slab models, wherein the total cost of the output slab model is lower than the total costs of other slab models.

The effect of the present invention is that the computer program product can automatically divide and flatten the sheet metal of the three-dimensional sheet metal model into different flattening schemes, and output the plate body model of the better flattening scheme, which effectively increases the efficiency of evaluating the sheet metal division of the three-dimensional sheet metal model. .

In order to describe the present invention more clearly, preferred embodiments are given and described in detail with the drawings as follows. Please refer to FIG. 1 , which is a structural diagram of a computer program product 100 according to a first preferred embodiment of the present invention. The computer program product 100 is used to evaluate the sheet metal division of the three-dimensional sheet metal model. The computer program product 100 is applied to the computer system 200 shown in FIG. 2 . The computer system 200 includes a host 201 , a screen 202 and an input device 203 connected to the host 201 . The host 201 is used for running the computer program product 100 , and displaying the picture through the screen 202 , and for the user to control through the input device 203 . The input device 203 may include a mouse 203a and a keyboard 203b. The host 201 can be a personal computer host, an industrial computer host, or a workstation host, etc., and has a storage medium for storing the computer program product 100 and three-dimensional graphics files.

The computer program product 100 includes a segmentation decision module 10 , a flattening decision module 20 and an evaluation module 30 .

The segmentation decision module 10 is used for receiving a three-dimensional image file, and the three-dimensional image file is stored in a storage medium. The three-dimensional drawing file includes a three-dimensional sheet metal model 300 of a sheet metal product (refer to FIG. 3 and FIG. 4 ), the three-dimensional sheet metal model 300 includes a plurality of spliced plate parts, and the three-dimensional sheet metal model 300 is mainly spliced by these plate parts made. The three-dimensional sheet metal model 300 as shown in FIG. 3 and FIG. 4 is composed of a plurality of sheet metal parts 310 . The sheet metal parts 310 include a main body sheet metal part 312 and a plurality of elongated sheet metal parts 314 , and each sheet metal part 310 has a plurality of each plate portion 310a. For example, in FIG. 4, the elongated sheet metal piece 314 has three plate portions 314a, 314b, 314c connected in sequence, and the plate portion 314b and each of its adjacent plate portions 314a, 314c have a common edge. The two plate portions 314a, 314c of the elongated sheet metal piece 314 are adjacent to the plate portion 312a of the main sheet metal piece 312, but have no common edge. All adjacent two plate portions 310a of the main body sheet metal piece 312 have a common edge between them.

The segmentation decision module 10 finds a plurality of the plate parts 310a of the three-dimensional sheet metal model 300 as a plurality of initial plate parts, and adds other plates adjacent to the initial plate parts from the initial plate parts respectively part to generate a plurality of three-dimensional sub-plate gold models. The generated three-dimensional sub-sheet metal models can be assembled into the three-dimensional sheet metal model 300 . The division decision module 10 treats each plate portion as a surface, and does not consider the thickness of each plate portion.

In this embodiment, the segmentation decision module 10 includes an object disassembly sub-module 12 , an initial board decision sub-module 14 , and a three-dimensional object generation sub-module 16 .

The three-dimensional sheet metal model 300 may include at least one object (not shown) coupled to at least one of the plate portions 310a, and the object may be, for example, a coupling element (such as a bolt, a rivet, a positioning pin) for coupling the two plate portions 310a. etc.), and/or the object may also be a sheet metal part with a size smaller than a predetermined size, or a non-sheet metal part (eg handle, circuit board, etc.). The aforementioned objects are likely to affect the judgment of the initial plate portion. Therefore, at least one object other than the plate portion 310a is removed from the three-dimensional sheet metal model 300 by the object dismantling sub-module 12, and only the plurality of plate portions 310a are retained, thereby avoiding errors in judging the initial plate portion.

The object dismantling sub-module 12 determines whether the object overlaps with the adjacent plate portion 310a, and if so, separates the overlap between the object and the plate portion 310a, separates the object and stores it in an object data set, and uses the three-dimensional plate model 300 to remove objects. The three-dimensional sheet metal model 300 having only the plate portion 310a is subjected to subsequent processing.

The initial board part decision sub-module 14 first selects at least one board part 310a from among the board parts 310a as at least one first initial board part, and the initial board part decision sub-module 14 can also select a plurality of board parts 310a as a plurality of first initial plate parts.

In this embodiment, the initial board part decision sub-module 14 analyzes the vertices of the board parts 310a, and calculates the number of adjacent edges of each vertex, and the number of adjacent edges is the degree of the vertex, please refer to FIG. 4 , the number of adjacent edges of vertex a of this long sheet metal piece is 3, and the number of adjacent edges of vertex b is 2. The initial board part decision sub-module 14 selects at least one board part from the board parts 310a that is connected with the vertex whose number of adjacent edges is greater than a predetermined value as the at least one first initial board part, or selects the largest number of adjacent edges The at least one plate portion connected to the vertices of the at least one first initial plate portion is used as the at least one first initial plate portion. The number of the first initial plate parts selected according to the aforementioned decision conditions may be one or more. The initial board part decision sub-module 14 stores the first initial board part in an initial board part data set.

For convenience of description, one first initial plate portion is used as an example for description, and it is assumed that the selected first initial plate portion is the plate portion 314b of the elongated sheet metal piece 314 (refer to FIG. 4 ). The 3D object generation sub-module 16 reads the initial plate part data set to form a three-dimensional sub-plate model with the first initial plate part. In the other surrounding plate parts 310a, select other plate parts 310a that can be flattened after adding the three-dimensional sub-plate model corresponding to the first initial plate part (that is, select other plate parts that are adjacent to each other and have a common edge), and add them to the corresponding plate parts 310a. In the 3D sub-plate model of the first initial plate part. In this way, a three-dimensional sub-plate model corresponding to the first initial plate portion can be generated.

When the three-dimensional object generation sub-module 16 determines that the selected plate part is added and the three-dimensional sub-plate model corresponding to the first initial plate part cannot be flattened, the initial plate part decision sub-module 14 sets the selected plate part 310a is at least one second initial board part and is stored in the initial board part data set. For example, when it is determined that the elongated sheet metal piece 314 and the board portion 312a of the main sheet metal piece 312 as shown in FIG. 3 are adjacent but do not have a common edge, the initial board portion decision sub-module 14 will The plate portion 312a of the main body sheet metal 312 is set as the second initial plate portion. The 3D object generation sub-module 16 reads the initial plate part data set to form another three-dimensional sub-plate model with the second initial plate part. Other plate parts 310a around the plate part that can be flattened after adding another three-dimensional sub-plate model are selected and added to the three-dimensional sub-plate model corresponding to the second initial plate part. In this way, a three-dimensional sub-plate model corresponding to the second initial plate portion can be generated.

Afterwards, if the three-dimensional object generating sub-module 16 determines that there is no other plate portion 310a available for generating the three-dimensional sub-sheet metal model, the three-dimensional sub-sheet metal model of all the sheet metal parts 310 is completed.

If the three-dimensional object generation sub-module 16 determines that the selected plate part 310a is added and the three-dimensional sub-plate model corresponding to the second initial plate part cannot be flattened, the initial plate part decision sub-module 14 will make the selected plate part To set another second initial board and save it to the initial board data set. Next, another new three-dimensional sub-plate model is generated as in the above-mentioned steps of generating the three-dimensional sub-plate model of the second initial plate portion. Until all the three-dimensional sub-plate models of the sheet metal parts 310 are generated. The initial board parts in the final initial board part data set include at least one first initial board part and at least one second initial board part.

The segmentation decision module 10 may further include a model detection sub-module (not shown in the figure), the model detection sub-module is used to debug each three-dimensional sub-plate model, and then debug the three-dimensional sub-plate model and the three-dimensional sub-plate model. The plate model 300 is compared to confirm whether the plate part of the 3D sub-plate model is the same as the plate part 310a of the corresponding plate part 310 in the 3-D plate model 300 . If it is different, it means that the 3D sub-plate model is wrongly generated, then the first initial plate part starts to generate all the 3-D sub-plate models, or a part of the 3-D sub-plate model is regenerated from the corresponding second initial plate part.

Then, the segmentation decision module 10 outputs all the three-dimensional sub-plate models to the flattening decision module 20 .

The flattening decision module 20 receives the three-dimensional sub-sheet models, and flattens the three-dimensional sub-sheet models to generate a plurality of flattening schemes, each of which includes a plurality of flattened sub-sheet models, each of which is flattened. Each of the flattened sub-sheet models in the scheme corresponds to each of the three-dimensional sub-sheet models, wherein at least a part of the three-dimensional sub-sheet models are flattened in different ways to form flattened sub-sheet models in different flattening schemes. In this embodiment, the flattening decision module 20 uses an algorithm to flatten each three-dimensional sub-plate model. The algorithm used may be a brute force algorithm (exhaustive search). All flat patterns are generated, among which, some 3D board models may have only one flattened pattern. For example, a 3D board model with only a bending structure can only be developed into a flattened sub-plate model. Some of the three-dimensional board models may have various flattened forms, for example, a three-dimensional board model including a structure in which three or more board parts are adjacent to each other can be developed into multiple flattened sub-sheet models. In addition, the flattening decision module records the cutting edges, bending edges, welding edges and corresponding side lengths required for processing each flattening sub-sheet model in each flattening sub-sheet model.

For example, in Table 1 below, the 3D sub-sheet model 1 can only be flattened into one flattened sub-sheet model 1, and the 3-D sub-sheet model 2 can be flattened in different ways to be flattened into the flattened sub-sheet model 1 and the flattened sub-sheet model. 2. The flattening decision module 20 forms a flattening scheme by arranging various combinations of possible flattening sub-plate models. When there are more possible flattened sub-sheet models for each three-dimensional sub-sheet model, there are more flattening schemes.

Table 1 Flattening Scheme 3D daughter plate gold model 1 3D daughter plate gold model 2 Flatten Scenario 1 Flattened sub-plate model 1 Flattened sub-plate model 1 Flatten Scenario 2 Flattened sub-plate model 1 Flattened Subplate Model 2

The evaluation module 30 receives the flattening schemes, and arranges the flattened sub-plate models in each of the flattening schemes on a plate body model. In this embodiment, the thickness of each plate portion is the same as an example, so the evaluation module 30 can arrange the flattened sub-plate models with the same thickness on the same plate body model. In practice, if the multiple flattened sub-sheet models in the flattening scheme have different thicknesses, the flattened sub-sheet models with the same thickness are arranged in the same plate model to form multiple plate models.

Please refer to FIG. 5 , which is an example in which a flattened sub-plate model 410 is arranged on a plate body model 400 in a flattening scheme. The evaluation module 30 arranges at least two flattened sub-plate models adjacent to each other, so that they have A common machined edge 420 . The processing edge is used for cutting, for example, the cutting line of laser cutting is the processing edge 420 . When there is a common processing edge 420, the number of cuttings can be reduced, so as to reduce the process cost.

Each flattening scheme has corresponding at least one plate body model. The evaluation module 30 calculates a total cost corresponding to each of the flattening schemes according to each of the plate models and a production cost parameter, and outputs at least one of the plate models, wherein one of the output plate models is The total cost is lower than that of other slab models.

In this embodiment, the production cost parameter includes a material cost parameter and a process cost parameter, the evaluation module calculates a material cost according to each of the plate body models and the material cost parameter, and calculates a material cost according to each of the plate body models and the process The cost parameter calculates the cost of an operation, and the total cost of each flattening plan is at least the sum of the material cost and the operation cost.

Material cost parameters include the price per unit area of the sheet. Sheet metal products have many choices in the sheet part, such as polished steel sheet, galvanized iron sheet, stainless steel sheet...etc. Different materials have different prices per unit area. The result of the above arrangement can determine the total area of materials required for a flattening scheme, and multiply the total required material area by the price per unit area of the used plate to obtain the material cost of the flattening scheme.

In addition, the process cost is related to the process used in production, such as cutting cost, folding bed, welding cost, etc. The process cost parameters include the unit price of the process used, including the price per unit length of laser cutting, the price per knife used by the folding machine, and the price per unit length of welding.

First of all, the sheet metal of the flattened flat sub-sheet metal model is cut from the sheet, and laser cutting is used. The cost of laser cutting is related to the thickness of the sheet and the length of the laser. Therefore, the process cost of the laser cutting process can be obtained by deducting the length of the common processing edge from the total perimeter of the cut edges of the flattened sub-plate models after being arranged, and multiplying it by the unit length price of laser cutting.

The number of times the folding machine is used is usually calculated by knives. The cost of using the folding bed is related to the number of discounts (ie the number of knives) and the length of the folding edge for each discount. The process cost of using the folding machine is obtained according to the number and length of the bent edges of the flattened sub-plate model and the price per knife.

In addition, the two edges after bending with a bending machine may need to be welded. The price per unit length of welding is related to the thickness of the plate, the material of the plate and the length of the welding, according to the length of the welded edge of the flattened sub-plate model and the unit of welding The process cost of welding can be obtained from the length price.

The evaluation module 30 adds up the material cost and all the process costs to obtain the total cost of each plate model. The evaluation module 30 can output the plate body model of the flattening scheme with the lowest total cost, or output the plate body model of the flattening scheme with the total cost lower than a predetermined cost, or output the material cost according to the level of the output material or the process cost. Plate model corresponding to high and low output.

The screening method of the plate body model of the output flattening scheme can be set by the user, or preset by the evaluation module 30 .

In a computer program product of a second preferred embodiment of the present invention, the structure is substantially the same as that of the first embodiment. The difference is that in this embodiment, the initial board decision sub-module determines the first initial board The method is to analyze the vertices of the plate parts 310a of the three-dimensional sheet metal model 300, and select the plate part 310a with the shortest relative distance between the vertices among the plate parts 310a or the plate part 310a with the smallest area formed between the vertices as the at least a first initial plate portion. That is, the smallest plate portion is selected as the first initial plate portion.

In the computer program product of a third preferred embodiment of the present invention, the structure is substantially the same as that of the first embodiment. The difference is that in this embodiment, the initial board decision sub-module determines the first initial board The method is to use a trained artificial neural network (ANN) to analyze the three-dimensional sheet metal model 300, and select at least one sheet portion 310a from the sheet portions 310a as at least a first initial sheet portion.

In the training process of training artificial neural network, since the number of neurons in the input layer is fixed, and the sizes of different 3D sheet metal models used for training are different, it is necessary to convert all training 3D sheet metal Model example for resizing. For example, as shown in FIG. 6 , a sphere 510 with a fixed resolution is selected, and the projection of the three-dimensional sheet metal model 500 on the sphere is used as the input data of the ANN, and the resolution of the inner surface of the sphere is high enough for digital quantization (quantization). , the appearance features of the 3D sheet metal model paradigm can be captured and can be applied to the input layer with a fixed number of neurons.

In another embodiment, a fixed resolution can be selected for the sphere, and the three-dimensional sheet metal model sample can be appropriately scaled and placed in the sphere and the aforementioned projection can be performed, which can also solve the aforementioned input. The data size is not fixed.

In another embodiment, the image information after the 3D sheet metal model is projected to the spherical surface can also be expanded into a flat surface through other algorithms, and then used as the input of the aforementioned ANN. These flat image data can also be Solve the aforementioned problem that the input data size is not fixed. In another embodiment, the image of the equidistant spherical surface can be projected on a rectangular plane through the equirectangular projection method, and the image information on the rectangular plane can be used as the image information of the aforementioned ANN. enter.

In another embodiment, the image of the equidistant sphere can be projected onto the surface of a cube through a cube map projection method. Then, the image information on the six square surfaces of the projected cube can be used as the input of the aforementioned ANN.

In another embodiment, the equi-angular cubemaps projection method can be used instead of the aforementioned cubemap projection method, and the image information on the six square surfaces of the projected cube can be used as the aforementioned ANN input of.

The aforementioned ANN must contain multiple hidden layers to capture the rules for selecting the initial board used by the segmentation decision module implicit in the training data as input to the ANN. Possible implementations of ANN include MLP (Multi-Layer Perceptron), CNN (Convolutional Neural Network, convolutional neural network) and other types of deep learning ANN.

The ANN training method is as follows:

1. Collect a collection of 3D sheet metal model examples of a certain type of sheet metal product;

2. For each 3D sheet metal model example in the above set, plus one or more manually marked initial sheet parts as labels;

3. In the way of supervised learning, use the labeled 3D sheet metal model example in the above set as a training set to train ANN. The training process may be repeated in multiple epochs. The 3D sheet metal model examples and labels within the ensemble are used to train the ANN until the output error of the ANN is less than an acceptable range.

The segmentation decision module uses ANN to find the first initial plate. The advantage is that it can cooperate with different types of sheet metal products and use the most appropriate method of selecting the initial plate to conduct the training of the segmentation decision module to select the first initial plate. It means that a specific ANN can be trained for a three-dimensional sheet metal model of a specific type of sheet metal product, and the ANN can be optimized by training the method of selecting the first initial plate part for the three-dimensional sheet metal model of this type of sheet metal product.

In this way, when faced with 3D sheet metal models of different types of sheet metal products, the segmentation decision module can select a trained ANN according to the type of sheet metal products, and capture the 3D sheet metal model in the projection method of the aforementioned corresponding training ANN. The appearance feature is applied to the fixed number of neurons in the input layer, that is, the selection of the first initial board can be performed.

According to the above, the computer program product of the present invention can divide and flatten the sheet metal of the three-dimensional sheet metal model into different flattening schemes, and output the plate body model of the better flattening scheme, thereby effectively increasing the efficiency of evaluating the three-dimensional sheet metal model.

The above descriptions are only preferred feasible embodiments of the present invention, and any equivalent changes made by applying the description of the present invention and the scope of the patent application should be included in the patent scope of the present invention.

100: Computer Program Products 10: Segmentation decision module 12: Object disassembly submodule 14: Initial board decision sub-module 16: 3D object generation sub-module 20: Flattening Decision Mods 30: Evaluation Module 200: Computer Systems 201: Host 202: Screen 203: Input device 203a: Mouse 203b: Keyboard 300: 3D Sheet Metal Model 310: Sheet metal parts 310a: Board Department 312: Main body sheet metal parts 312a: Board Department 314: Long sheet metal pieces 314a: Board Department 314b: Board Department 314c: Board Department a: vertex b: vertex 400: Plate Model 410: Flatten the daughter plate gold model 420: machined edge 500: Examples of 3D Sheet Metal Models 510: Sphere

FIG. 1 is a structural diagram of a computer program product according to a first preferred embodiment of the present invention. FIG. 2 is a computer system to which the computer program product of the first preferred embodiment of the present invention is applied. FIG. 3 is a three-dimensional sheet metal model of the first preferred embodiment of the present invention. FIG. 4 is an elongated sheet metal part of a three-dimensional sheet metal model according to the first preferred embodiment of the present invention. FIG. 5 is a partial schematic view of the plate body model of the first preferred embodiment of the present invention. FIG. 6 is a schematic diagram of projecting a three-dimensional sheet metal model template for training onto a sphere according to the third preferred embodiment of the present invention.

100: Computer Program Products

10: Segmentation decision module

12: Object disassembly submodule

14: Initial board decision sub-module

16: 3D object generation sub-module

20: Flattening Decision Mods

30: Evaluation Module

Claims (10)

A computer program product for sheet metal division for evaluating three-dimensional sheet metal models, comprising: a division decision module for receiving a three-dimensional drawing file, the three-dimensional drawing file includes a three-dimensional sheet metal model, and the three-dimensional sheet metal model includes a plurality of splices the board parts, the segmentation decision module finds a plurality of the board parts in the three-dimensional sheet metal model as a plurality of initial board parts, and respectively adds the other board parts adjacent to the initial board parts from the initial board parts The plate part is used to generate a plurality of three-dimensional sub-plate models, wherein these three-dimensional sub-plate models can be spliced into the three-dimensional plate model; a flattening decision module receives the three-dimensional sub-plate models, and converts the three-dimensional sub-plate models into Flattening the sheet metal model to generate a plurality of flattening schemes, each of the flattening schemes includes a plurality of flattened sub-sheet models, and each of the flattened sub-sheet models in each of the flattened schemes corresponds to each of the three-dimensional sub-sheet models, wherein at least A part of the three-dimensional sub-plate model includes a structure in which more than three plate parts are adjacent to each other and has a variety of flattened patterns, so as to form a flattened sub-plate model in different flattening schemes; and an evaluation module receives these Flatten the plan, and arrange the flattened sub-plate models in each flattening plan on a plate body model, and the evaluation module calculates a corresponding flattening plan according to each of the plate body models and a production cost parameter. and outputting at least one of the plate body models, wherein the total cost of the output plate body model is lower than the total cost of the other plate body models. The computer program product for evaluating sheet metal segmentation of a three-dimensional sheet metal model as claimed in claim 1, wherein the initial plate parts include at least a first initial plate part and at least a second initial plate part, and the segmentation decision module is set in At least one of the plate parts of the three-dimensional sheet metal model is selected as the at least one first initial plate part, and the at least one first initial plate part is used as the basis of at least one of the three-dimensional sub-plate models, and then the at least one Others around the first initial plate Other plate parts that can still be flattened after the three-dimensional sub-plate model corresponding to the at least one first initial plate part are selected from the plate part, and added to the three-dimensional sub-plate model corresponding to the at least one first initial plate part; when When the segmentation decision module determines that the selected board part is added and the three-dimensional sub-plate model corresponding to the at least one first initial board part cannot be flattened, the selected board part is set as the at least one second initial board part, And use the at least one second initial plate part as the basis of another three-dimensional sub-plate model, and then select and add another three-dimensional sub-plate model from other plate parts around the at least one second initial plate part. The other plate parts are flat and added to the three-dimensional sub-plate model corresponding to the at least one second initial plate part. The computer program product for evaluating the sheet metal segmentation of a three-dimensional sheet metal model as claimed in claim 2, wherein the segmentation decision module includes an initial plate decision sub-module, and the initial plate decision sub-module is used to determine the At least one plate portion is selected as the at least one first initial plate portion. The computer program product for evaluating sheet metal segmentation of a three-dimensional sheet metal model as described in claim 3, wherein the initial board decision sub-module analyzes vertices of the board parts, and selects relative distances between the vertices among the board parts The shortest board part or the board part with the smallest area formed between the vertexes is used as the at least one first initial board part. The computer program product for evaluating the sheet metal segmentation of a three-dimensional sheet metal model as described in claim 3, wherein the initial board part decision sub-module analyzes the vertices of the board parts, and calculates the number of adjacent edges of each vertex, and selects The at least one board part connected to the vertex with the number of adjacent edges greater than a predetermined value is selected as the at least one first initial board part, or at least one board part connected to the vertex with the largest number of adjacent edges is selected as the at least one first initial board part board. The computer program product for evaluating sheet metal segmentation of a three-dimensional sheet metal model as described in claim 3, wherein the initial board decision sub-module uses a trained artificial neural network (Artificial Neural Network, ANN), at least one board part is selected from the board parts as the at least one first initial board part. The computer program product for evaluating the sheet metal division of a three-dimensional sheet metal model as claimed in claim 3, wherein the initial board part decision sub-module selects the initial board parts, and stores the initial board parts in an initial board part data set ; Wherein the segmentation decision module includes a three-dimensional object generation sub-module, the three-dimensional object generation sub-module reads the initial plate part data set, and adds the obtained initial plate parts adjacent to each initial plate part Connect other plate parts with common edges to generate each of the three-dimensional sub-plate models. The computer program product for evaluating the sheet metal division of a three-dimensional sheet metal model according to claim 1, wherein when the evaluation module arranges the flattened sub-sheet metal models in each of the flattening schemes on a plate body model, at least two The flattened sub-plate models are arranged adjacent to each other so that they have a common processing edge. The computer program product for evaluating the sheet metal division of a three-dimensional sheet metal model as described in claim 1, wherein the production cost parameter includes a material cost parameter and a process cost parameter, and the evaluation module is based on each of the plate body models and the material cost The parameter calculates a material cost, and calculates a process cost according to each of the plate body models and the process cost parameter, and the total cost of each of the flattening schemes is at least the sum of the material cost and the process cost. The computer program product for evaluating sheet metal division of a three-dimensional sheet metal model as claimed in claim 1, wherein the division decision module includes an object disassembly sub-module, and the three-dimensional sheet metal model includes at least one object combined with at least one of the plate parts; the The object disassembly submodule is used for removing the at least one object from the three-dimensional sheet metal model.

Images