KR20230010500A - Apparatus and method for predicting and providing surface roughness of product to be moulded, and apparatus and method for predicting and providing process condition, using artificial intelligence - Google Patents

Abstract

The present invention relates to an apparatus and a method for predicting and providing surface roughness of a molded product and an apparatus and a method for predicting and providing molding conditions by using artificial intelligence (AI). The apparatus for predicting and providing surface roughness of a molded product by using AI according to the present invention comprises: a display; a memory storing matching information on matched images of surface roughness and molding conditions, which is generated by using AI; an input module where molding conditions for molding a product are inputted; and a processor outputting an expected image of surface roughness of a molded product to be molded according to the received molding conditions through the display, based on the matching information.

Description

Apparatus and method for predicting and providing the surface roughness of a sculpture using artificial intelligence, and apparatus and method for predicting and providing molding conditions FOR PREDICTING AND PROVIDING PROCESS CONDITION, USING ARTIFICIAL INTELLIGENCE}

The present invention relates to a device and method for predicting and providing the surface roughness of a sculpture using artificial intelligence, and a device and method for predicting and providing molding conditions.

Recently, interest in three-dimensional (three-dimension) printing is increasing. For example, interest in 3D printing using directed energy deposition is increasing. This is because the direct energy stacking method facilitates high-quality part repair and large-scale part manufacturing.

However, the surface state (surface roughness or surface roughness) of an object molded using the direct energy deposition method may vary depending on the process condition. For example, a sculpted object formed through the direct energy deposition method may have an uneven surface state. Accordingly, post-processing (eg, surface processing) of the sculpture may be required, which may waste time, manpower, and/or cost.

The present invention is to solve the above-mentioned problems of the prior art, the present invention is to predict (predict) the surface roughness of the object to be molded by the input molding conditions, and use artificial intelligence that can provide it to improve the surface roughness of the object Its purpose is to provide a device and method for predicting and providing.

In addition, the present invention learns an artificial intelligence (AI) model based on the learning data of a pre-shaped sculpture, and the optimal modeling conditions capable of generating a sculpture having a desired surface roughness are learned through the AI model. An object of the present invention is to provide a device and method for predicting and providing modeling conditions using artificial intelligence that can be predicted and provided.

An apparatus for predicting and providing surface roughness of a sculpture using artificial intelligence according to an embodiment of the present invention for achieving the above object includes a display; a memory that is generated using the artificial intelligence and stores matching information in which a surface roughness image and a molding condition are matched; An input module for inputting molding conditions for modeling a sculpture; And based on the matching information, it may include a processor for outputting an expected surface roughness image of a sculpture to be shaped according to the received molding condition through the display.

A method for predicting and providing surface roughness of a sculpture using artificial intelligence according to an embodiment of the present invention includes receiving molding conditions for modeling a sculpture through an input module; and providing an expected surface roughness image of a sculpture to be molded according to the received molding condition based on matching information generated using the artificial intelligence and stored in a memory and matching the surface roughness image and the molding condition. can do.

An apparatus for predicting and providing molding conditions of a molding apparatus using artificial intelligence according to an embodiment of the present invention includes a memory for storing a learned artificial intelligence model; And obtaining a reference image having a desired surface roughness, preprocessing the acquired reference image, and analyzing the preprocessed reference image through the artificial intelligence model to form a sculpture having a surface roughness corresponding to the reference image. It may include a processor that predicts and provides molding conditions.

A method for predicting and providing molding conditions of a molding apparatus using artificial intelligence according to an embodiment of the present invention includes obtaining a reference image having a desired surface roughness; pre-processing the obtained reference image through an image processing module; and analyzing the preprocessed reference image through an artificial intelligence model to predict and provide modeling conditions for modeling a sculpture having a surface roughness corresponding to the reference image.

As described above, the present invention can check the surface roughness of a sculpture to be shaped in advance according to the input molding conditions, and can provide an effect of reducing time, cost, and / or manpower for manufacturing a sculpture.

In addition, the present invention can easily obtain (e.g., recommended by AI) optimal modeling conditions for modeling a sculpture having a desired surface roughness, through a molding device (e.g., a 3D printing device). It is possible to provide an effect of reducing time, cost, and/or manpower for manufacturing a sculpture.

1 is a diagram showing a system for modeling a sculpture according to an embodiment of the present invention.
2 is a block diagram showing the configuration of the control device of FIG. 1;
3 is a flowchart illustrating a learning operation of artificial intelligence according to an embodiment of the present invention.
4 is an exemplary diagram for explaining a method of acquiring learning data for artificial intelligence learning according to an embodiment of the present invention.
5 is a flowchart illustrating a method of recommending optimal molding conditions using artificial intelligence according to an embodiment of the present invention.
6 is an exemplary diagram for explaining a method of recommending optimal modeling conditions using artificial intelligence according to an embodiment of the present invention.
7 is a diagram for explaining a method of providing an estimated surface roughness image according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. 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, but will be implemented in various different forms, only these 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 fully inform the holder of the scope of the invention, and the present invention is only defined by the scope of the claims. In the following, like reference numerals designate like components.

Although first, second, etc. are used to describe various elements, components and/or sections, it is needless to say that these elements, components and/or sections are not limited by these terms. These terms are only used to distinguish one element, component or section from another element, component or section. Accordingly, it goes without saying that the first element, first element, or first section referred to below may also be a second element, second element, or second section within the spirit of the present invention.

Terminology used herein is for describing the embodiments and is not intended to limit the present invention. In this specification, singular forms also include plural forms unless specifically stated otherwise in a phrase. As used in the specification, a referenced component, step, operation and/or element to "comprises" and/or "made of" refers to one or more other components, steps, operations and/or elements. Existence or additions are not excluded.

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.

Hereinafter, the configuration of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram showing a system for modeling a sculpture according to an embodiment of the present invention.

Referring to FIG. 1 , a system according to an embodiment of the present invention may include a molding device 100 and a control device 200 .

The molding device 100 may mold a sculpture. For example, the molding device 100 may be a 3D printing device. Hereinafter, a case where the molding apparatus 100 is a 3D printing apparatus using a direct energy deposition method will be described as an example.

The molding device 100 may mold a sculpture according to a set molding condition (process condition). The molding conditions may include laser power (intensity), powder supply speed, and scan speed. The molding apparatus 100 moves the nozzle along a designated pattern at a set speed, discharges powder (eg, Ti-6Al-4V alloy powder) at a set powder supply speed, and combines the powders emitted by the laser with the set power. (e.g. deposited in a layered manner).

The control device 200 may be wired or wirelessly connected to the molding device 100 to control the molding device 100 . For example, the control device 200 may be a desktop PC or laptop PC that can be connected to the molding device 100 by wire or wirelessly. As another example, the control device 200 may be a server. As another example, the control device 200 may be included in the molding device 100 .

The control device 200 trains and learns based on learning data (including a learning image and modeling conditions) related to a prefabricated object, and optimizes the object for modeling a object having a surface roughness desired by the user. It may include an artificial intelligence (AI) model that predicts and provides molding conditions of A detailed description thereof will be described later with reference to FIGS. 3 to 6 .

Also, when molding conditions are input, the control device 200 may predict and provide the surface roughness of the object to be molded according to the input molding conditions. A detailed description thereof will be described later with reference to FIG. 8 .

2 is a block diagram showing the configuration of the control device of FIG. 1;

Referring to FIG. 2 , the control device 200 may include a processor 210, a memory 220, a display 230, an input module 240, and a scan module 250. The memory 220 may include an image processing module 221 , a first AI model 223 and a second AI model 225 .

The processor 210 may perform various functions by receiving commands or instructions from the memory 220 and controlling each component according to the received commands or instructions. The processor 210 according to the present invention provides a procedure for providing optimal modeling conditions for modeling a sculpture having a desired surface roughness through an AI model and a procedure for predicting and providing the surface roughness of a sculpture to be shaped according to the modeling conditions. You can control it. A detailed description of the processor 210 controlling each procedure will be described later with reference to FIGS. 5 to 7 .

The processor 210 may be formed of a central processing unit (CPU), a micro control unit (MCU), a micro processor unit (MPU), or the like.

The memory 220 may be electrically connected to the processor 210, and the memory 220 contains various programs for operating the control device 200 and various commands and/or instructions for operating the processor 210. can be saved The memory 220 may include at least one of an internal memory and an external memory. The built-in memory includes volatile memory (eg, DRAM, SRAM, SDRAM, etc.), non-volatile memory (eg, OTPROM (one time programmable ROM), PROM, EPROM, EEPROM, mask ROM, flash ROM, flash memory, hard drive, Or it may include at least one of a solid state drive (SSD) The external memory may include SD (secure digital), Micro-SD, Mini-SD, or USB (secure digital) that may be connected to the control device 200 through various interfaces. universal serial bus) memory, etc.

The memory 220 may include an image processing module 221 , a first AI model 223 , and a second AI model 225 .

The image processing module 221 may support image processing of a target image for predicting a learning image for learning of the first AI model 223 and a modeling condition. Also, the image processing module 221 may support a data augmentation function. 2 shows that the image processing module 221 is a software module stored in the memory 220 . However, according to another example, the image processing module 221 may be a hardware module included in the processor 210 or a hardware module included in the control device as a separate component.

The first AI model 223 may predict and provide optimal modeling conditions for modeling a sculpture having a surface roughness desired by a user. To this end, the first AI model 223 may perform training and learning based on learning data related to prefabricated sculptures. The learning data may include a scanned image of a sculpture (hereinafter referred to as a learning image) and modeling conditions for modeling the object.

The first AI model 223 may be a deep convolution neural network. However, the present invention is not limited to the first AI model 223 being a deep convolutional neural network, and various known artificial intelligence techniques may be used.

The second AI model 225 may generate a virtual image similar to a designated reference image (hereinafter referred to as a real image). For example, the second AI model 225 may generate a virtual surface roughness image based on images augmented by the image processing module 221 . The second AI model 225 may be generative adversarial networks (GAN) or deep convolution generative adversarial networks (DCGAN). However, the present invention is not limited to that the second AI model 225 is GAN or DCGAN, and various known artificial intelligence techniques may be used.

Virtual surface roughness images generated by the second AI model 225 may be input to the first AI model 223 to predict modeling conditions. The predicted modeling condition may be matched with a corresponding virtual surface roughness image and stored (eg, stored as matching information). Alternatively, the predicted molding condition may be labeled on the corresponding virtual surface roughness image.

The display 230 may provide an input function and/or an output function. For example, the display 230 may include a touch panel and/or a display panel. The display 230 may display an optimal modeling condition predicted by the first AI model 223 . Also, the display 230 may display an expected surface roughness image of an object to be modeled according to input modeling conditions.

The input module 240 may include a physical button, an optical key, a keypad, a mouse, or a touch pad. The input module 240 may receive a user input for inputting molding conditions. When the input module 240 is formed as a touch panel, the input module 240 may be integrally formed with the display 230 .

The scan module 250 may scan the sculpture. The scan module 250 may generate a surface roughness image (hereinafter referred to as a scan image) by scanning the sculpture. The scanned image may be used as training data for learning of the first AI model 223 or may be used as a reference image for the second AI model 225 to generate a surface roughness image of lyrics. Meanwhile, the scan module 250 may not be included in the control device 200 but may be included in the molding device 100 or may be a separate component.

3 is a flowchart for explaining a learning operation of artificial intelligence according to an embodiment of the present invention, and FIG. 4 is a flowchart for explaining a method of acquiring learning data for learning of artificial intelligence according to an embodiment of the present invention. is an example

Referring to FIGS. 3 and 4 , the control device 200 (eg, the processor 210) according to an embodiment of the present invention may obtain learning data (S301). The learning data may include learning images 421 and 423 obtained by scanning the sculpture 410 pre-shaped through the modeling device 100 in 2D or 3D and modeling conditions for modeling the corresponding sculpture 410. Hereinafter, description will be made using the training image 421 scanned in two dimensions.

The control device 200 may pre-process the training image 421 (S303). For example, the control device 200, as shown in the identification code 430 of FIG. 4, cuts out the blank of the training image 421, and as shown in the identification code 440 of FIG. 4, the specified size (eg : 128*128) can be resized.

The control device 200 may augment the preprocessed training image (data augmentation) (S305). For example, the control device 200 rotates (451, 453, 455) the resized image by a specified angle (eg, 90 degrees, 180 degrees, 207 degrees), as shown by the identification numeral 450 in FIG. 4 . , it is possible to generate new learning images by horizontally inverting (457) or vertically inverting (459). Meanwhile, step S305 is for preventing overfitting, and may be omitted if training data is sufficient. Although not shown, the control device 200 may perform normalization to adjust the RGB values of the training image to a value between 0 and 1.

The control device 200 may perform learning of the first AI model 223 using the learning image (S307). The first AI model 223 may include a deep convolution neural network. However, the present invention is not limited to the first AI model 223 being a deep convolutional neural network, and various known artificial intelligence techniques may be used.

5 is a flowchart illustrating a method of recommending optimal modeling conditions using artificial intelligence according to an embodiment of the present invention, and FIG. 6 is an optimal modeling condition using artificial intelligence according to an embodiment of the present invention. It is an exemplary diagram for explaining a method of recommending a condition.

Referring to FIGS. 5 and 6 , the control device 200 may receive a reference image 610 (S501). The reference image 610 may be a scanned image of a sculpture having a surface roughness desired by the user.

The control device 200 may pre-process the reference image 601 (S503). For example, as shown in 603 of FIG. 6 , the control device 200 may rescale the reference image 610 to reduce capacity and improve efficiency.

The control device 200 may analyze the preprocessed reference image (S505). For example, as shown in 605 of FIG. 6 , the control device 200 may use an AI model (e.g., : The reference image preprocessed through the first AI model 223) may be analyzed. The AI model may be a deep convolutional neural network (DCNN), as shown at identification 605 . A deep convolutional neural network may consist of one input layer, one output layer, and four hidden layers. The activation function of the hidden layer may include Relu, and the activation function of the output layer may include sigmoid. Meanwhile, the structure of the deep convolutional neural network of identification code 605 is only an example and does not limit the present invention.

The control device 200 may provide predicted optimal molding conditions (S507). For example, the control device 200 may output the predicted optimal molding condition on the display 230 as indicated by identification code 607 .

Meanwhile, the control device 200 automatically or manually (eg, a user's request) transmits the calculated modeling conditions to the modeling device 100 so that the modeling device 100 can mold a sculpture according to the corresponding modeling conditions. .

On the other hand, as a result of evaluating the performance of the first AI model by predicting the modeling conditions of the trained image set and the non-learning image set (test set) through the pre-learned first AI model, the prediction that predicted the modeling condition Predicted data and original data representing actual molding conditions were similar. In detail, referring to the evaluation result shown by identification code 611, the mean absolute percentage error (MAPE) of the laser power for all the evaluated training images is 3.28%, and the mean absolute percentage error of the powder supply speed is 3.28%. It can be seen that the error is 5.11% and the mean absolute error of the progress rate is 1.89%. In addition, referring to the evaluation result shown by identification code 613, the average absolute ratio error of the laser power for the entire non-learning image evaluated is 7.81%, the average absolute ratio error of the powder supply speed is 11.77%, and the progress It can be seen that the average absolute percentage error of the speed is 4.30%. This is just an example, and more accurate prediction will be possible if the first AI model learns through a larger number of training images. Prediction results for non-learning images will also become more accurate.

7 is a diagram for explaining a method of providing an estimated surface roughness image according to an embodiment of the present invention.

First, the control device according to an embodiment of the present invention may further include a second AI model generating a virtual image.

Referring to FIG. 7 , the control device according to an embodiment of the present invention may scan a sculpture and pre-process the scanned image (S701). The control device may perform data augmentation by rotating or inverting the preprocessed images (S703). Data augmented images may be input to the second AI model 720 . The second AI model 720 may include a generator 721 that generates a fake image and a discriminator 722 that determines the generated fake image.

The second AI model 720 may generate virtual surface roughness images based on the data augmented images (S705). The generated virtual surface roughness images may have saturation and brightness adjusted (S707).

The virtual surface roughness images generated by the second AI model 720 are input to the first AI model 710, and modeling conditions may be respectively predicted (S709). The predicted modeling conditions may be matched with virtual surface roughness images and stored. Through this, the control device of the present invention can secure (eg, store in memory) matching information (eg, a database or matching table) in which various molding conditions and surface roughness images are matched.

When molding conditions are input through the input module, the control device (eg, the processor of the control device) according to an embodiment of the present invention described above uses matching information to match the surface roughness image (expected surface roughness image) with the molding condition. ) can be provided. For example, the control device may extract a surface roughness image matching the input modeling condition from matching information, and output the extracted surface roughness image on a display. Through this, the user can check the surface roughness of the object to be modeled in advance without modeling the object through the modeling device.

Although the above has been described with reference to the illustrated embodiments of the present invention, these are only examples, and those skilled in the art to which the present invention belongs can variously It will be apparent that various other embodiments that are variations, modifications and equivalents are possible. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

100: molding device 200: control device
210: processor 220: memory
221: image processing module 222: first AI model
223: second AI model 230: display
240: input module 250: scan module

Claims (13)

In the apparatus for predicting and providing the surface roughness of a sculpture using artificial intelligence,
display;
a memory that is generated using the artificial intelligence and stores matching information in which a surface roughness image and a molding condition are matched;
An input module for inputting molding conditions for modeling a sculpture; and
Based on the matching information, the apparatus characterized in that it comprises a processor for outputting an expected surface roughness image of the object to be shaped by the received modeling condition through the display.
According to claim 1,
The processor
preprocessing at least one training image;
Rotate or flip the preprocessed training image to perform data augmentation,
Through a second artificial intelligence model, based on the augmented learning images, a virtual surface roughness image is generated,
Analyzing the generated virtual surface roughness image through a first artificial intelligence model to predict molding conditions for modeling a sculpture corresponding to the generated virtual surface roughness image, and
The apparatus characterized in that the virtual surface roughness image and the predicted modeling conditions are matched and stored as the matching information.
In the method of predicting and providing the surface roughness of a sculpture using artificial intelligence,
Receiving modeling conditions for modeling a sculpture through an input module; and
Providing an expected surface roughness image of a sculpture to be shaped by the received molding condition based on matching information generated using the artificial intelligence and stored in a memory and matching the surface roughness image and the molding condition characterized by a method.
According to claim 3,
pre-processing at least one training image;
Rotating or flipping the preprocessed training image to perform data augmentation;
generating a virtual surface roughness image based on the augmented learning images through a second artificial intelligence model;
Predicting modeling conditions for modeling a sculpture corresponding to the generated virtual surface roughness image by analyzing the generated virtual surface roughness image through a first artificial intelligence model; and
The method further comprising the step of matching the virtual surface roughness image and the predicted modeling condition and storing the matching information.
A device for predicting and providing molding conditions of a molding device using artificial intelligence,
a memory to store the learned artificial intelligence model; and
Obtaining a reference image having a desired surface roughness, preprocessing the acquired reference image, and analyzing the preprocessed reference image through the artificial intelligence model to form a sculpture having a surface roughness corresponding to the reference image. A device comprising a processor that predicts and provides conditions.
According to claim 5,
The processor
and performing the preprocessing by rescaling the reference image.
According to claim 5,
The device further comprising a display outputting the molding conditions.
According to claim 5,
The artificial intelligence model is
An apparatus comprising a deep convolution neural network.
According to claim 5,
The processor
Acquiring learning data including modeling conditions and learning images;
Pre-processing the training image in a designated manner,
Rotate or flip the preprocessed training image to perform data augmentation,
The device characterized in that for performing the learning of the artificial intelligence model through the data-augmented learning image.
According to claim 5,
The molding device is a 3D printing device of a directed energy deposition method,
The molding conditions include laser power, powder supply speed, and scan speed of the three-dimensional printing device.
A method for predicting and providing molding conditions of a molding apparatus using artificial intelligence,
obtaining a reference image having a desired surface roughness;
pre-processing the obtained reference image through an image processing module; and
Analyzing the preprocessed reference image through an artificial intelligence model, predicting and providing modeling conditions for modeling a sculpture having a surface roughness corresponding to the reference image.
According to claim 11,
The step of pre-processing the obtained reference image
and rescaling the reference image to a specified size.
According to claim 11,
Acquiring learning data including modeling conditions and learning images;
pre-processing the training image;
Rotating or flipping the preprocessed training image to perform data augmentation; and
The method further comprising the step of performing learning of the artificial intelligence model through the data augmented training images.

Images