KR20230061845A - Method and apparatus for predicting CAE result data of injection molding of automotive parts - Google Patents

Abstract

Disclosed are a method and an apparatus for predicting the computer-aided engineering (CAE) result of automobile parts. The method for predicting the CAE result of automobile parts according to one embodiment of the present invention comprises the steps of: allowing a collection unit to collect a data set including first data which are 3D data of the automobile parts and second data which are the CAE result data obtained by performing CAE on the 3D data as initial data necessary for predicting the CAE result of automobile parts; allowing a conversion unit to convert the file format of the initial data such that the first data and the second data can be compared one-to-one; allowing a learning data set generation unit to generate a learning data set from the initial data whose file format has been converted; allowing an artificial intelligence model of a prediction unit to predict the CAE result data corresponding to the modeled data for any automobile part when the artificial intelligence model of the prediction unit is trained with the learning data set; and allowing a visualization unit to visualize the modeled data and the predicted CAE result data. Accordingly, even in the early stages of designing automobile parts, CAE results, such as deformation after injection, can be checked to reduce the human and material resources required to obtain final design data.

Description

Method and apparatus for predicting CAE result data of injection molding of automotive parts}

A method and apparatus for predicting CAE results of automotive parts are disclosed.

Automotive parts are completed by repeating processes such as design, skin data generation, 3D data generation, and part performance analysis. Specifically, after completing the design of the exterior of automobile parts, skin data is created by 3D modeling the exterior shape, and then, based on the skin data, design work is carried out to add the fastening structure and rigidity securing structure of the parts. Generate 3D data. Then, the performance of the parts is verified using these 3D data. If the performance verification result by 3D numerical analysis is unsatisfactory, the exterior design of the automobile part is changed, and the skin data creation, 3D data creation, and parts performance verification processes are repeated.

The steps as described above are repeatedly performed until the performance verification result reaches the target value. Therefore, there is a problem in that considerable time and cost are required to acquire the final design data of automobile parts.

Republic of Korea Patent No. 10-2287659 (Title of Invention: Defect Prediction System and Method for Plastic Injection Molding Process, Date of Publication: August 9, 2020)

A method and apparatus for predicting CAE results of automotive parts are disclosed.

In order to solve the above problems, a method for predicting CAE results of automobile parts according to an embodiment includes first data, which is 3D data of the automobile parts, and the collecting a data set including second data that is CAE result data obtained by performing CAE on 3D data; converting the file format of the initial data so that the conversion unit can compare the first data and the second data one-to-one; generating a training data set from the initial data of which the file format is converted by a training data set generating unit; predicting CAE result data corresponding to data modeled for an arbitrary automobile part using the learned artificial intelligence model when the artificial intelligence model of the prediction unit is learned from the training data set; and visualizing the modeled data and the predicted CAE result data by a visualization unit.

In order to solve the above problems, an apparatus for predicting a CAE result of an automobile part according to an embodiment includes first data that is 3D data of the automobile part and the 3D data as initial data necessary to predict the CAE result of the automobile part. a collection unit that collects a data set including second data that is CAE result data obtained by performing CAE on the target; a converter converting a file format of the initial data so that the first data and the second data can be compared one-to-one; a training data set generating unit generating a training data set from the initial data whose file format is converted; a prediction unit that predicts CAE result data corresponding to data modeled for an arbitrary automobile part using an artificial intelligence model learned from the learning data set; and a visualization unit that visualizes the modeled data and the predicted CAE result data.

Since CAE results, for example, post-injection deformation can be checked even in the initial stage of designing automobile parts, human and material resources required to obtain the final design data of automobile parts can be reduced.

1 is a diagram showing the configuration of an apparatus for predicting CAE results of automobile parts according to an embodiment of the present invention.
2 is a diagram illustrating a visualization of an example of a data set having a standard file format.
FIG. 3 is a diagram illustrating a comparison between a part of data before file format conversion and a part of data after file format conversion in relation to the data set shown in FIG. 2 .
4 is a diagram illustrating a method of selecting an adjacent node based on a distance to a reference node.
5 is a diagram illustrating a part of a training data set.
6 is a diagram illustrating a state in which 3D data, which is input data of the prediction unit, and CAE result data, which are output data of the prediction unit, are visualized.
7 is a hardware configuration diagram of an exemplary computing device capable of implementing a prediction device according to an embodiment of the present invention.
8 is a flowchart illustrating a method of predicting a CAE result of an automobile part according to an embodiment of the present invention.

Advantages and features of the present invention, and methods of achieving them, will become clear with reference to the detailed description of the following embodiments taken in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms, only the present embodiments make the disclosure of the present invention complete, and common knowledge in the art to which the present invention belongs. It is provided to completely inform the person who has the scope of the invention, and the present invention is only defined by the scope of the claims.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used in a meaning commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless explicitly specifically defined.

Terminology used herein is for describing the embodiments and is not intended to limit the present invention. In this specification, the singular form also includes the plural form unless otherwise specified in the entry and exit phrase. As used herein, "comprises" and/or "comprising" does not exclude the presence or addition of one or more other elements other than the recited elements.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, like reference numerals denote like elements.

1 is a diagram showing the configuration of a device for predicting a CAE result of an automobile part (hereinafter, referred to as a 'prediction device') according to an embodiment of the present invention.

Referring to FIG. 1, the prediction device 100 includes a collection unit 110, a conversion unit 120, a training data set generation unit 130, a database 140, an input unit 150, a prediction unit 160, and a visualization unit. It includes section 170.

The collection unit 110 collects initial data required to predict CAE results of automobile parts. The initial data to be collected may be a data set including first data and second data.

It may be understood that the first data is 3D data for automobile parts. 3D data is the skin data of automotive parts created by various types of CAD (Computer-Aided Design) / CAM (Computer-Aided Manufacturing) / CAE (Computer Aided Engineering) software. It can be obtained by carrying out design work that adds a fastening structure and rigidity securing structure of parts based on skin and vertex weight). . The second data may be understood as CAE result data obtained by performing CAE on 3D data of automobile parts. As CAE result data, injection molding results for automobile parts are exemplified. The injection molding result may include, for example, information on injection molding deformation, in particular, information on post-deformation after injection molding.

The collection unit 110 may collect data sets for each vehicle part. The collected data set may have a file format before being converted to a standard file format, or may have a standard file format. Examples of standard file formats include the STL file format and the IGES file format.

STL (STereoLithography) is a file format of stereo-lithography CAD software, and is a kind of polygon format that expresses the surface of a three-dimensional object, that is, a 3D model composed of countless triangular faces. Since the STL file format is composed of triangles, it is difficult to express curved surfaces, but if you increase the number of divisions of triangles and draw them into more delicate triangles, you can get a shape almost similar to a curved surface. In addition to 'STereoLithography', the abbreviation STL is also used as 'Standard Transformation Language', 'Surface Tesselation Language', and 'Standard Triangulation Language'.

IGES (Initial Graphics Exchange Specification) is a standard established by the National Bureau of Standards (NBS) of the US Department of Commerce for the exchange of graphic information in 1980. Hereinafter, IGES will be described in more detail.

IGES aims to provide a neutral data format for digital representation and communication of product definition data. IGES is the most widely used among many neutral file formats developed in many countries around the world, so it can be said to be a de facto world standard. Since version 1.0 of IGES was released in 1980, version 6.0 including solid models was released in 1998.

The basic unit of data expressed in IGES is called an entity, and supported entities can be largely classified into three types: Geometry Entity, Annotation Entity, and Structure Entity.

Geometry entities represent curves, surfaces, and solids, and annotation entities represent annotation-related information such as leaders, dimensions, symbols, and cross-sections. In addition, the structure entity represents the size of a character, the thickness of a line, the definition of a color, a grouping relationship, a finite element model (FEM) element, and the like.

The IGES file format is largely classified into ASCII format and binary format.

ASCII format is divided into fixed line length format and compressed format. The fixed line length format is the most commonly used format, and data is recorded in units of lines consisting of 80 columns, but the file size is large. The compressed ASCII format has the advantage of being able to reduce the size of IGES files created in fixed line length format when the size is an issue. The reason is that the compressed ASCII format can eliminate redundant data by using a notation format that incorporates some paragraphs in the fixed line length format.

Binary format is ASCII format represented as a binary bit string. The binary format consists of a Byte-Oriented structure suitable for transmission of large files. Referring back to FIG. 1 , the conversion unit 120 converts the file format of the data set collected by the collection unit 110 .

For example, if the file format of the collected data set is not the same, the conversion unit 120 converts the file format of the data set into one of the standard file formats described above. The standard file format to convert the collected data set to can be set by the user in advance, and the set value can be implemented to be changeable. In the following description, a case in which the conversion unit 120 converts the collected data set into an STL file format will be described as an example.

As another example, if the file format of the collected data set is a standard file format, the conversion unit 120 converts the file format of the data set into a file format for injection molding analysis, for example, a Moldflow ^® UDM file format. Moldflow ^® is software for plastic injection molding simulation. Here, with reference to FIGS. 2 and 3, the conversion to a file format for injection molding analysis will be described in more detail.

2 is a diagram illustrating a visualization of an example of a data set having a standard file format. FIG. 3 is a diagram illustrating a comparison between a part of data before file format conversion and a part of data after file format conversion in relation to the data set shown in FIG. 2 .

As shown in Fig. 2, the standard file format, i.e., the STL file format, is composed of triangles. However, since the STL file format consists of only the X, Y, and Z coordinate values for each vertex of a triangle, it is impossible to compare 3D data included in the data set and CAE result data on a 1:1 basis. By the way, if the STL file format is converted to the UDM file format, it is possible to compare the 3D data included in the data set and the CAE result data on a 1:1 basis.

In (A) of FIG. 3, a part of data in the STL file format is exemplified, and in (B) of FIG. 3, a part of data in the UDM file format is exemplified.

Referring to (A) of FIG. 3 , it can be seen that the data in the STL file format includes only the X, Y, and Z coordinate values of each vertex of the triangle. In contrast, referring to (B) of FIG. 3 , it can be seen that each triangle is identified by an element number and the vertices of each triangle are identified by a node number in the UDM file format data. Also, it can be seen that each node includes X, Y, and Z coordinate values. Referring to (B) of FIG. 3 , it can be seen that a triangle having an element number of 8 includes vertices having node numbers of 1135, 1163, and 1164, respectively. As such, in the UDM file format, since triangles and their vertices can be identified by element numbers and node numbers, respectively, it is possible to compare 3D data included in the data set with CAE result data on a 1:1 basis.

Referring back to FIG. 1 , the training data set generator 130 generates a training data set based on the data set converted by the conversion unit 120 . To this end, the training data set generator 130 may perform an operation on the converted data set or extract data from the converted data set. For example, data conversion such as standardization and normalization may be performed, or data extraction such as stratified extraction and weighted stratified extraction may be performed. Standardization expresses how far the data is from the mean by dividing the data by the standard deviation, and examples include log transformation and root transformation. Normalization refers to setting the maximum and minimum values and transforming the data into quantiles within the range. Stratified extraction refers to dividing data into multiple regions according to a distribution and extracting data at the same rate for each divided region. Weighted stratification extraction refers to dividing data into a plurality of regions according to the distribution and extracting random rate data for each divided region. Hereinafter, a process of generating a training data set will be described in more detail.

First, the training data set generator 130 compares nodes corresponding to each other in the converted first data and the converted second data, and calculates a change amount for each node. Since the node includes X, Y, and Z coordinate values, the amount of change in the X coordinate value, the amount of change in the Y coordinate value, and the amount of change in the Z coordinate value between the corresponding nodes are calculated, respectively.

Thereafter, the training data set generator 130 selects adjacent nodes corresponding to the reference number for each node. The reference number may be, for example, 500. However, the reference number is not limited to the exemplified ones, and may be less than 500 or more than 500. The reference number may be adjusted according to the prediction performance of the artificial intelligence model included in the prediction unit 160 to be described later.

When a node selected from among nodes included in the first data is referred to as a 'reference node', adjacent nodes may be determined based on a distance to the reference node. That is, among the nodes located around the reference node, nodes corresponding to 500 may be determined as adjacent nodes in order of distance from the reference node. 4 illustrates a method of selecting an adjacent node based on a distance to a reference node.

When the adjacent nodes are determined, the training data set generating unit 130 calculates the amount of change of the reference node for the adjacent nodes, respectively. That is, the amount of change in X coordinate values, the amount of change in Y coordinate values, and the amount of change in Z coordinate values between the reference node and a specific adjacent node are calculated, respectively.

Meanwhile, values obtained through the above-described calculation process may be recorded in a data sheet for a corresponding node. Here, the data sheet will be described in more detail with reference to FIG. 5 .

5 is a diagram illustrating data sheets for some nodes.

Referring to FIG. 5 , data sheets for node 1, node 2, node 3, node 4, and node N are illustrated. Hereinafter, the data sheet for node 1 will be mainly described.

Referring to FIG. 5, the data sheet for node 1 includes input_node, outpu_node, input_x, input_y, input_z, output_x, output_y, output_z, move_x, move_y, move_z, e, input_x_re, input_y_re, and input_z_re items.

input_node is a part where a node number in the first data is recorded. The output_node is a part where a node number in the second data is recorded. Node numbers recorded in input_node and output_node are the same.

input_x, input_y, and input_z are portions where X coordinate values, Y coordinate values, and Z coordinate values of corresponding nodes in the first data are respectively recorded.

output_x, output_y, and output_z are parts in which the X coordinate value, Y coordinate value, and Z coordinate value of the corresponding node are respectively recorded in the second data.

move_x, move_y, and move_z are portions in which changes in X coordinate values, Y coordinate values, and Z coordinate values are respectively recorded.

input_x_re, input_y_re, and input_z_re are the parts where the amount of change in X coordinate value, Y coordinate value change amount, and Z coordinate value change amount of node 1 with respect to the adjacent node are respectively recorded.

Referring to the data sheet for node 1, it can be seen that the data sheet includes data for node 1 and data for nodes adjacent to node 1.

Data sheets as shown in FIG. 5 may constitute a data set for learning. A data set for training may be stored in the database 140 . The stored learning data set may be used to learn an artificial intelligence model of the prediction unit 160 to be described later.

Referring back to FIG. 1 , the prediction unit 160 predicts a CAE result by receiving 3D data of an arbitrary automobile part. To this end, the prediction unit 160 may include an artificial intelligence model. The artificial intelligence model may be implemented as one of various types of artificial neural network models. Artificial neural network models include Deep Feedforward Network (DFN), Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), Deep Residual Network (DRN), Examples include Long Short-Term Memory (LSTM) neural networks, Bidirectional Long Short-Term Memory (biLSTM) neural networks, and transformer neural networks.

The artificial intelligence model of the prediction unit 160 may be implemented as a bidirectional LSTM neural network among the artificial neural network models illustrated. Whereas LSTM neural networks contain only one LSTM layer, bidirectional LSTM neural networks contain two LSTM layers. Here, one LSTM layer uses the input in the forward direction, and the other LSTM layer uses the input in the backward direction. Bidirectional LSTM neural networks have higher prediction accuracy than LSTM neural networks because they learn in both directions rather than unidirectionally.

The prediction unit 160 as described above may be trained using a training data set stored in the database 140 .

The input unit 150 receives modeled data for a predetermined automobile part, for example, an automobile part for which a CAE result is to be predicted. Examples of the modeled data include skin data and 3D data. Skin data is data obtained by 3D modeling the exterior shape of automobile parts. 3D data is data obtained by adding parts fastening structures and rigidity securing structures based on skin data. Hereinafter, for convenience of explanation, a case in which 3D data is received as modeling data will be described as an example.

In this way, to receive modeled data of automobile parts, the input unit 150 may include a user input unit for receiving information from a user. The user input unit may include, for example, a touch key and a mechanical key. The 3D data input through the input unit 150 is provided to the prediction unit 160 and the visualization unit 170 to be described later.

The prediction unit 160 inputs the 3D data provided from the input unit 150 to the artificial intelligence model. As a result, the artificial intelligence model outputs CAE result data for the input 3D data. The CAE result data may include an injection molding result, and the injection molding result may include information about a degree of deformation after injection molding. The prediction unit 160 transfers the CAE result data output from the artificial intelligence model to the visualization unit 170.

The visualization unit 170 visualizes the CAE results provided from the prediction unit 160 . Specifically, the visualization unit 170 may display the CAE result data provided from the predictor 160 by overlapping the 3D data provided from the input unit 150 . At this time, the 3D data and the CAE result data may be displayed in different colors. For a more detailed description of this, reference will be made to FIG. 6 .

6 is a diagram illustrating visualization of 3D data, which is input data of an artificial intelligence model, and CAE result data, which is output data of an artificial intelligence model.

Referring to FIG. 6 , it can be seen that 3D data, which is the input data of the artificial intelligence model, is displayed in green. In addition, it can be seen that the CAE result data, which is the output data of the artificial intelligence model, is marked in red.

6 shows a case where the output data of the artificial intelligence model is displayed in red, but according to another embodiment, the output data may be displayed in a plurality of colors according to the degree of deformation after injection. More specifically, the output data may be displayed in the order of blue, green, yellow, orange, and red as the degree of deformation after injection increases.

By visualizing the input data and output data of the artificial intelligence model in different colors, users can intuitively recognize the relationship between 3D data and CAE result data. In addition, with reference to the recognized result, the design of 3D data for automobile parts may be changed.

Meanwhile, each component of the prediction device 100 shown in FIG. 1 may mean software or hardware such as a field programmable gate array (FPGA) or an application-specific integrated circuit (ASIC). . However, the components are not limited to software or hardware, and may be configured to be in an addressable storage medium or configured to execute one or more processors. Functions provided within the components may be implemented by more subdivided components, or may be implemented as a single component that performs a specific function by combining a plurality of components.

For reference, FIG. 1 shows that the prediction device 100 is a single physical device, but this is only for convenience of understanding, and according to an embodiment, the prediction device 100 is composed of a plurality of computing devices. It can also be implemented as a system. For example, one or more of the collection unit 110, the conversion unit 120, the learning data generation unit 130, the prediction unit 140, and the visualization unit 170 may be implemented as an independent computing device. In this case, each computing device may communicate through a network.

In the above, the configuration and operation of the prediction device 100 according to an embodiment of the present invention have been described with reference to FIGS. 1 to 6 . Hereinafter, an exemplary computing device capable of implementing the prediction device 100 according to some embodiments of the present invention will be described with reference to FIG. 7 .

7 is a hardware configuration diagram of an exemplary computing device in which devices in accordance with some embodiments of the invention may be implemented.

Referring to FIG. 7 , a computing device 200 includes one or more processors 210, a storage 290 for storing a computer program 251, and a computer program 251 executed by the processor 210 loaded. It may include a memory 220, a bus 230, and a network interface 240. However, only components related to the embodiment of the present invention are shown in FIG. 7 . Therefore, those skilled in the art to which the present invention pertains can know that other general-purpose components may be further included in addition to the components shown in FIG. 7 .

The processor 210 controls the overall operation of each component of the computing device 200 . The processor 210 includes a Central Processing Unit (CPU), a Micro Processor Unit (MPU), a Micro Controller Unit (MCU), a Graphic Processing Unit (GPU), or any type of processor well known in the art. It can be. Also, the processor 210 may perform an operation for at least one application or program for executing a method according to embodiments of the present invention. Computing device 200 may include one or more processors.

The memory 220 stores one or more of various data, commands and information. The memory 220 may load one or more programs 251 from the storage 250 in order to execute the learning data generating method according to embodiments of the present invention. In FIG. 7 , RAM is illustrated as an example of the memory 220 .

The bus 230 provides a communication function between components of the computing device 200 . The bus 230 may be implemented as various types of buses such as an address bus, a data bus, and a control bus.

The network interface 240 supports wired and wireless Internet communication of the computing device 200 . Also, the network interface 240 may support various communication methods other than Internet communication. To this end, the network interface 240 may include a communication module well known in the art.

Storage 250 may non-temporarily store one or more computer programs 291 . The storage 250 may be a non-volatile memory such as read only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, or the like, a hard disk, a removable disk, or a device well known in the art. It may be configured to include any known type of computer-readable recording medium.

8 is a flowchart illustrating a method of predicting a CAE result of an automobile part according to an embodiment of the present invention.

First, the prediction device 100 collects initial data necessary for predicting CAE results of automobile parts (S710). The initial data to be collected may be a data set including first data and second data. In addition, the first data may be understood as 3D data of automobile parts, and the second data may be understood as CAE result data obtained by performing CAE on the 3D data.

Then, the prediction device 100 converts the file format of the collected initial data (S720). Step S720 includes, when the file format of the collected initial data is not a standard file format, converting the file format of the collected initial data into a standard file format (eg, STL file format), and the collected initial data If the file format of is a standard file format, converting the file format of the collected initial data into a file format for CAE analysis (eg, UDM file format) is included. When the step S720 is completed, a one-to-one comparison between nodes included in the first data and nodes included in the second data is possible.

Thereafter, the prediction device 100 generates a training data set from the initial data of the converted file format (S730). The step S730 includes the steps of comparing nodes corresponding to each other in the converted first data and the converted second data and calculating the amount of change for each node, a reference node selected from the converted first data and an adjacent node of the reference node. Calculating the amount of change between each node, and generating a data sheet in which the calculated values are recorded for each node. A data sheet generated for each node may constitute a data set for learning.

Thereafter, the prediction device 100 learns the artificial intelligence model of the prediction unit 160 using the learning data set generated in step S730 (S740). The artificial intelligence model can be implemented as a bidirectional LSTM neural network, for example.

Thereafter, the prediction device 100 predicts CAE result data for 3D data of a predetermined vehicle part by using the artificial intelligence model that has been learned (S750). That is, the AI model that has been trained takes 3D data of automobile parts as input data and outputs CAE result data for the 3D data as output data corresponding to the input data.

Thereafter, the prediction device 100 visualizes 3D data, which is input data of the artificial intelligence model, and CAE result data, which is output data (S760). The step S760 may include overlapping input data with output data and displaying the input data and output data in different colors.

A result of visualizing the input data and the output data may be output through an output unit (not shown) such as a display (S770).

In the above, the embodiments of the present invention have been described. In addition to the foregoing embodiments, the embodiments of the present invention may be implemented through a medium containing computer readable code/instructions for controlling at least one processing element of the foregoing embodiment, for example, a computer readable medium. there is. The medium may correspond to a medium/media enabling storage and/or transmission of the computer readable code.

The computer readable code may not only be recorded on a medium, but may also be transmitted through the Internet. The medium is, for example, a recording medium such as a magnetic storage medium (eg, ROM, floppy disk, hard disk, etc.) and an optical recording medium (eg, CD-ROM, Blu-Ray, DVD), a carrier wave ( carrier wave). Since the medium may be a distributed network, computer readable code may be stored/transmitted and executed in a distributed manner. Still further, by way of example only, a processing element may include a processor or a computer processor, and the processing element may be distributed and/or included within a device.

Although the embodiments of the present invention have been described with reference to the exemplified drawings as described above, those skilled in the art to which the present invention pertains may change the technical spirit or essential features of the present invention in other specific forms. It will be appreciated that this can be implemented. Therefore, the embodiments described above are illustrative in all respects, and should be understood as non-limiting.

100: prediction device
200: computing device

Claims (11)

A data set including first data, which is 3D data of the automobile part, and second data, which is CAE result data obtained by performing CAE on the 3D data, as initial data required for the collection unit to predict the CAE result of the automobile part collecting;
converting the file format of the initial data so that the conversion unit can compare the first data and the second data one-to-one;
generating a training data set from the initial data of which the file format is converted by a training data set generating unit;
predicting CAE result data corresponding to data modeled for an arbitrary automobile part using the learned artificial intelligence model when the artificial intelligence model of the prediction unit is learned from the training data set; and
Visualizing the modeled data and the predicted CAE result data by a visualization unit,
How to predict CAE results for automotive parts. According to claim 1,
The step of converting the file format of the initial data,
converting the file format of the initial data into a standard file format when the file format of the initial data is not a standard file format; and
When the file format of the initial data is a standard file format, converting the file format of the initial data so that the first data and the second data can be compared one-to-one.
How to predict CAE results for automotive parts. According to claim 1,
The file format of the initial data in which the file format is converted is a UDM file format,
How to predict CAE results for automotive parts. According to claim 3,
The step of generating the training data set,
calculating a first change amount between nodes corresponding to each other in the first data and the second data of the converted initial data;
calculating second variations between a reference node selected from the first data and nodes adjacent to the reference node, respectively;
generating a data sheet including the first change amount and the second change amount; and
Generating the data sheet for all nodes included in the first data,
How to predict CAE results for automotive parts. According to claim 1,
The data modeled for the arbitrary automobile parts includes skin data and 3D data,
The skin data is data obtained by 3D modeling the exterior shape of the automobile part,
The 3D data is data obtained by adding a fastening structure and a rigidity securing structure of parts based on the skin data,
How to predict CAE results for automotive parts. According to claim 1,
The AI model is
Including a bidirectional long short-term memory (biLSTM) neural network,
How to predict CAE results for automotive parts. According to claim 1,
The predicted CAE result data includes injection molding results for the modeled data,
The injection molding result includes information on post-deformation after injection molding,
How to predict CAE results for automotive parts. According to claim 1,
The visualization step is
superimposing output data of the artificial intelligence model on input data of the artificial intelligence model;
displaying the input data in a first color; and
Including; displaying the output data in a second color,
The input data is 3D data of the arbitrary automobile parts,
The output data is the predicted CAE result data,
How to predict CAE results for automotive parts. According to claim 8,
In the step of displaying the output data in a second color,
configuring the second color to include a plurality of colors;
applying the plurality of colors differently according to the predicted CAE result data; and
Including the step of displaying the output data in the different applied colors,
How to predict CAE results for automotive parts. Collecting a data set including first data, which is 3D data of the automobile part, and second data, which is CAE result data obtained by performing CAE on the 3D data, as initial data necessary to predict the CAE result of the automobile part collection department;
a converter converting a file format of the initial data so that the first data and the second data can be compared one-to-one;
a training data set generating unit generating a training data set from the initial data whose file format is converted;
a prediction unit that predicts CAE result data corresponding to data modeled for an arbitrary automobile part using an artificial intelligence model learned from the training data set; and
A visualization unit for visualizing the modeled data and the predicted CAE result data; including,
A device that predicts CAE results of automotive parts. Combined with a computing device,
Collecting a data set including first data, which is 3D data of the automobile part, and second data, which is CAE result data obtained by performing CAE on the 3D data, as initial data necessary to predict the CAE result of the automobile part step;
converting a file format of the initial data so that the first data and the second data can be compared one-to-one;
generating a training data set from the initial data whose file format is converted;
predicting CAE result data corresponding to data modeled for an arbitrary automobile part by using an artificial intelligence model that has been learned from the training data set; and
A computer program implemented to perform the step of visualizing the 3D data of the arbitrary automobile part and the predicted CAE result data, stored in a computer-readable recording medium.

Images