Method and system for automating a metal processing operation to minimize burr formation and optimize machining efficiency using an artificial intelligence model trained based on metal processing data

Abstract

The present invention relates to a method and system for automating a metal processing process to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data. A system for automating a metal processing process to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data according to one embodiment of the present invention comprises: a memory; a communication unit; The device includes a processor connected to the communication unit and memory and executing program commands stored in the memory, wherein the processor collects metal processing process data in real time, including cutting condition information, tool status information, material information, and sensor data during processing from a metal processing device, and uses the collected metal processing process data as input values ​​to predict the probability of burr generation during processing as a probability value between 0 and 1 through a first model that has been trained, and based on the probability of burr generation, recalculates the probability of burr generation by resetting the value of at least one process variable among feed rate, spindle rotation speed, cutting depth, and cutting path as the input value of the first model, and obtains process conditions by repeatedly performing the change of process variables and prediction by the first model until the recalculated probability of burr generation becomes less than a preset reference value, transmits the process conditions to the metal processing device to adjust the processing conditions in real time, and can retrain the first model based on the metal processing process data additionally collected during the processing and the data obtained by applying the burr generation result received from the metal processing device or operator terminal after the processing is completed to the metal processing process data.

Description

Technology Field

[0001] The present invention relates to a method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data. More specifically, the invention relates to a method and system for automating the process flow from the drawing input stage to CAD design, CAM conversion, and CNC processing data generation when performing fiber laser processing and CNC processing on various metal plates such as steel plates, stainless steel plates, and aluminum plates, and to optimizing processing efficiency while minimizing the generation of burrs and dross by utilizing an artificial intelligence model learned based on metal processing process data.

[0002] In particular, the present invention relates to a method and system for automating a metal processing process to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned from metal processing process data, which performs intelligent process control technology that minimizes quality deviation and reduces post-processing by optimally controlling complex process variables such as laser output, processing speed, focal position, auxiliary gas conditions, material thickness, and material properties in real-time or based on prior prediction. Background Technology

[0004] Generally, the processing of metal sheet materials such as steel plates, stainless steel, and aluminum is carried out by receiving 2D or 3D drawings from a customer, creating or modifying drawings using a CAD program, setting tool paths and processing conditions through CAM software, generating CNC code (G-code, etc.), and performing actual processing using a fiber laser cutter or CNC equipment.

[0005] Furthermore, when multiple parts are placed on a single sheet metal for cutting, a nesting process is performed beforehand to properly arrange the part shapes and maximize raw material utilization efficiency. The nesting process is a key step for minimizing the loss rate of the original sheet metal and reducing scrap generation.

[0006] However, the following problems exist in conventional technology.

[0007] First, nesting work relies heavily on the worker's experience and skill level.

[0008] When arranging multiple parts of different shapes and sizes on a single sheet of steel, significant experience is required to determine the optimal arrangement to maximize material yield, and even the results of automatic arrangement frequently require manual correction by an operator.

[0009] Second, the nesting results are not considered integrally with actual processing conditions.

[0010] Although factors such as spacing between parts, cutting sequence, and heat concentration areas affect machining quality, conventional systems operate by separating layout optimization and machining condition settings.

[0011] Third, the setting of processing conditions relies on the operator's experience.

[0012] Settings such as laser output, cutting speed, auxiliary gas pressure, focal position, feed rate, gas type (nitrogen, oxygen, air cut), nozzle position, and frequency vary depending on the material and thickness; if these settings are improper, various cutting quality defects occur due to aging or malfunction of components such as lenses, nozzles, and nozzle adapters mounted on the laser cutting header.

[0013] In particular, in conventional metal processing processes, cutting quality defects frequently occur, such as drilling failures or defects at the starting point during piercing, uneven cutting surfaces depending on the cutting direction, incomplete cutting at the edge, adhesion of moss-like structures, slag, or burrs to the processed surface, formation of dross beneath the cut, deterioration of head or processing precision (head misalignment, processing deviation, etc.), processing instability and thermal deformation caused by changes in cutting speed, and incomplete cutting after piercing.

[0014] These defects are caused by complex factors such as minute deviations in process conditions, changes in equipment status, differences in material properties, and heat accumulation effects, and have the limitation of being difficult to predict quantitatively in advance.

[0015] Fourth, when burrs and dross occur, additional deburring or post-processing is required, which reduces productivity and increases manufacturing costs.

[0016] Fifth, the CAD-CAM-CNC conversion process is fragmented, leading to data compatibility issues and repetitive modification work, and requires resetting the previous step when process conditions change.

[0017] Sixth, even under identical material and thickness conditions, quality variations occur depending on equipment status, nozzle wear, lens contamination, and ambient temperature; however, there is a lack of intelligent systems that systematically learn these factors and incorporate them into the process.

[0018] Consequently, conventional technology operates nesting optimization, processing condition settings, quality prediction, and equipment control independently of one another, relying mostly on post-processing correction or experience-based responses. As a result, problems such as increased raw material loss rates, the occurrence of burrs and dross, unstable cutting quality, increased post-processing costs, and reduced productivity continue to occur. The problem to be solved

[0020] The present invention was devised to fundamentally resolve the problems of manual nesting dependency, experience dependency in setting process conditions, and poor cutting quality arising from the aforementioned conventional technology. It aims to provide a method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model trained on metal processing data capable of solving the following technical challenges.

[0021] First, the goal is to automate the nesting process of placing multiple parts onto a single metal plate and minimize raw material loss, while simultaneously providing an integrated nesting optimization structure that considers heat concentration and quality impacts resulting from placement positions and cutting sequences.

[0022] Second, by using an artificial intelligence model trained on metal processing data including cutting conditions, tool condition, material properties, equipment condition, and sensor data during processing, it is possible to predict in advance the possibility of cutting quality defects such as piercing defects, deterioration of cutting surface uniformity according to the cutting direction, edge failure, adhesion of slag and burrs to the processing surface, dross generation, deterioration of head precision, processing instability due to changes in cutting speed, and incomplete cutting after piercing, before or during processing.

[0023] Third, based on the above prediction results, an adaptive control structure is provided that automatically calculates multiple process conditions, including feed rate, spindle speed, cutting depth, laser output, focal position, auxiliary gas pressure, cutting path, and cutting sequence, and reflects them in real time to the equipment control unit, thereby suppressing the occurrence of the above defect in advance.

[0024] Fourth, it implements a self-improving process control system by feeding back process data collected in real-time during processing to an artificial intelligence model to correct prediction accuracy, and by reflecting even the quality inspection results after processing is completed as training data.

[0025] Fifth, the goal is to establish a data-driven prediction and control structure that differentiates itself from existing post-correction-centered metalworking methods by linking nesting optimization, defect prediction, automatic process condition adjustment, and equipment control into a single integrated platform. means of solving the problem

[0027] A method and system for automating a metal processing process to minimize burr generation and optimize processing efficiency using an artificial intelligence model learned based on metal processing process data according to an embodiment of the present invention comprises: a memory; a communication unit; The system includes a processor connected to the communication unit and memory to execute program commands stored in the memory, wherein the processor collects metal processing process data in real time from a metal processing device, including cutting condition information, tool status information, material information, and sensor data during processing, and using the collected metal processing process data as input values, predicts the probability of burr occurrence during processing as a probability value between 0 and 1 through a pre-trained first model, and based on the probability of burr occurrence, recalculates the probability of burr occurrence by resetting the value obtained by changing at least one process variable among feed rate, spindle speed, cutting depth, and cutting path as the input value of the first model, and obtains process conditions by repeatedly performing the change of process variables and prediction by the first model until the recalculated probability of burr occurrence becomes less than a pre-set reference value, transmits the process conditions to the metal processing device to adjust the processing conditions in real time, and retrains the first model based on the metal processing process data additionally collected during the processing and the data obtained by applying the burr occurrence result received from the metal processing device or operator terminal after the completion of processing to the metal processing process data, and the first model includes feed rate, spindle speed, cutting It may be an artificial intelligence model trained to determine the possibility of burr generation during machining by using process variables including depth and cutting path, tool condition, material properties, and sensor data as input data, and burr generation results confirmed after machining is completed as ground truth data.

[0028] According to one embodiment of the present invention, the processor, upon receiving cutting condition information including two-dimensional or three-dimensional shape data for a metal part from the metal processing device or the operator terminal, executes a nesting algorithm based on the two-dimensional or three-dimensional shape data for the metal part to determine the shape, production quantity, material, thickness, cutting interval, thermal stress distribution by cutting direction, residual heat accumulation effect, laser heat affected zone (HAZ) or heat distribution generated by the tool, minimum gap between neighboring metal parts in a batch, batch angle, cutting path, and statistical or simulation-based residual scrap generation probability of the metal part; based on the nesting algorithm, obtains recommended process data including a batch drawing of the metal part, cutting sequence, and cutting path; sets the recommended process data as input variables of the first model to obtain a prediction result of the probability of burr generation through the first model; updates the process conditions calculated based on the prediction result of the probability of burr generation based on equipment status information including nozzle wear, lens contamination, tool wear, and temperature change collected in real time from the metal processing device; and the recommended process data and the Updated process information can be provided to the metal processing device in real time.

[0029] A processor according to an embodiment of the present invention stores work data collected for each worker from the metal processing device or the worker terminal, executes a plurality of second models that have learned worker-specific characteristics using the work data as input data, executes a third model learned to determine a batch similarity category using two-dimensional or three-dimensional shape data of the metal part as input data, calculates a worker priority based on a selected category according to the output result of the third model and previously stored information on burr occurrence and material loss rate for each category, selects a second-1 model for a worker of a certain rank or higher corresponding to the selected category among the plurality of second models based on the priority, performs an ensemble operation by applying a first weight corresponding to the priority to the output value of each of the selected plurality of second-1 models, derives work data in which the possibility of burr occurrence is below a preset threshold and the material loss rate is minimized, corrects the recommended process data based on the derived work data, wherein the work data includes the worker's drawing modification history, setting values, the type of metal processing device used, the type of metal to be processed, the occurrence of burrs, and the material loss rate, and the setting values ​​are the metal processing At least one of the frequency of the output wave for the device, processing speed, assist gas pressure, focus position, and nozzle gap is a value adjusted by the operator, and the second model is an artificial intelligence model trained to determine the possibility of burr generation and material loss rate when a specific operator performs work under specific process conditions using the operator's work data and the burr generation result and material loss rate according to the work result as training data, and the third model may be an artificial intelligence model trained to determine a batch similar category corresponding to a target drawing using the shape and size of the workpiece, existing batch data, and process data as training data.

[0030] According to one embodiment of the present invention, the processor calculates a score for each worker with a rank of at least a certain level corresponding to the selected category based on the output value of each of the selected plurality of 2-1 models, sets a second weight for each worker for each of the selected plurality of 2-1 models based on the calculated score, and performs an ensemble operation by applying a first weight corresponding to the priority to the output value of each of the selected plurality of 2-1 models, wherein the ensemble operation is performed by additionally applying the second weight, and if a worker whose calculated score is below a preset threshold is identified, the 2-2 model corresponding to the identified worker among the selected plurality of 2-1 models is excluded from the selected plurality of 2-1 models and the 2-2 model is deleted, and if new work data is collected for the identified worker, the 2-3 model is created and retrained using the newly collected work data as training data for the identified worker, and the 2-3 model is formed using the work data of the worker and the burr generation result and material loss rate according to the work result as training data for the specific It may be an artificial intelligence model trained to determine the likelihood of burr generation and material loss rate when performing work under process conditions.

[0031] According to one embodiment of the present invention, the processor collects metal oxide data and vibration data in real time from a sensor device comprising at least one first sensor for detecting at least one metal oxide within a metal processing area of ​​the metal processing device and a second sensor for detecting vibration of a metal processing workbench; for each of the metal oxide data and the vibration data, performs a correction to remove the acquired data as noise data for a predetermined period of time from the start of processing of the metal processing device; determines an abnormal situation based on the degree of metal oxide generation and the degree of vibration generation based on the corrected data; if the result of the determination determines that a workpiece conforming to a predetermined defect criterion is generated, generates an alarm through the metal processing device or the operator terminal; and if a work stoppage command is received from the metal processing device or the operator terminal, provides a control command to the metal processing device to stop the operation of the metal processing device; stores the metal oxide data and the vibration data from the start of processing of the metal processing device to the stoppage of operation of the metal processing device as abnormal situation data; sets a work stoppage condition based on the degree of metal oxide generation and the level of vibration generation based on the accumulated abnormal situation data; and performs work in the work stoppage condition or the alarm generation situation. The amount of workpiece waste is obtained as work progress data from the above-mentioned operator terminal, and based on the above-mentioned work progress data, the average amount of workpiece waste is calculated according to the oxide generation concentration by metal type and the vibration frequency by metal type, and if the calculated average amount of workpiece waste exceeds a preset threshold ratio, a forced work stop value for the oxide generation concentration by metal type and the vibration frequency by metal type is set based on the above-mentioned work progress data to generate forced stop data, and the accumulated and stored forced work stop value,From a fourth model, which is an artificial intelligence model trained to determine work stoppage criteria and alarm response using the accumulated work progress data and corresponding waste amount data as training data, results are obtained regarding the determination of a forced stoppage situation and an alarm occurrence situation based on the metal oxide data and vibration data additionally collected during a future processing process; if a forced stoppage situation is determined during a future processing process based on the output of the fourth model, a control command to stop the operation of the metal processing device is provided to the metal processing device; and if an alarm occurrence situation is determined during a future processing process based on the output of the fourth model, an alarm can be generated through the metal processing device or the operator terminal.

[0032] A device according to one embodiment of the present invention may be combined with hardware and controlled by a computer program stored on a medium to execute any one of the operation methods by the processor of the system described above. Effects of the invention

[0034] According to the method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model trained on metal processing process data of the present invention, the following effects can be expected.

[0035] First, by predicting the possibility of eight types of cutting quality defects occurring before or during processing and automatically adjusting process conditions, it becomes possible to implement prevention-oriented quality control rather than post-processing correction.

[0036] Second, piercing defects, incomplete cutting, burrs, and dross are significantly reduced, minimizing deburring and reprocessing steps, which in turn reduces production lead time and post-processing costs.

[0037] Third, processing instability caused by heat concentration and variations in cutting surface quality depending on the cutting direction is reduced, thereby improving cutting surface uniformity and precision.

[0038] Fourth, by integrating and optimizing nesting results and processing conditions, it is possible to achieve comprehensive quality stabilization that reflects the correlation between batch, cutting sequence, and quality, going beyond simple material yield improvement.

[0039] Fifth, through real-time data-based feedback control, it can adaptively respond to changes in equipment status (nozzle wear, optical system contamination, tool wear, etc.), thereby maintaining quality stability even during repetitive production under the same conditions.

[0040] Sixth, as process data and quality inspection results are continuously learned, the prediction accuracy of the artificial intelligence model gradually improves, enabling the implementation of a self-evolving manufacturing system that actively responds to changes in the production environment.

[0041] Seventh, automatic nesting and automatic setting of process conditions are possible even in multi-product, small-batch production environments, significantly improving production flexibility and the level of automation.

[0042] Consequently, the present invention functions not as a simple processing automation system, but as a data-based intelligent metal processing integrated control platform that predicts and controls cutting quality defects in advance, and can provide technical effects that simultaneously achieve burr reduction and productivity improvement. Brief explanation of the drawing

[0044] FIG. 1 is a block diagram schematically illustrating the basic configuration of a metal processing automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data in one embodiment. FIGS. 2 and 3 are schematic diagrams illustrating a metal processing automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data according to one embodiment. FIGS. 4 and 5 are basic operation flowcharts for a metal processing process automation method of a metal processing process automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data according to one embodiment. Specific details for implementing the invention

[0045] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0046] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0047] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0048] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or joined to that other component, or that there may be other components in between.

[0049] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0050] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0051] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0052] The embodiments can be implemented in various forms of products such as personal computers, laptop computers, tablet computers, smartphones, televisions, smart home appliances, intelligent automobiles, kiosks, and wearable devices.

[0053] Artificial Intelligence (AI) systems are computer systems that implement human-level intelligence; unlike existing rule-based smart systems, they are systems in which machines learn and make decisions autonomously. As AI systems improve in recognition accuracy and gain a more accurate understanding of user preferences with continued use, existing rule-based smart systems are gradually being replaced by deep learning-based AI systems.

[0054] Artificial intelligence technology consists of machine learning and component technologies utilizing machine learning. Machine learning is an algorithmic technology that autonomously classifies and learns the characteristics of input data, while component technologies are technologies that mimic the cognitive and judgmental functions of the human brain by utilizing machine learning algorithms such as deep learning, and are comprised of technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control.

[0055] The various fields where artificial intelligence technology is applied are as follows. Linguistic understanding refers to technologies that recognize, apply, and process human language and text, including natural language processing, machine translation, dialogue systems, question answering, and speech recognition / synthesis. Visual understanding refers to technologies that perceive and process objects like human vision, including object recognition, object tracking, image search, person recognition, scene understanding, spatial understanding, and image enhancement. Inference and prediction refers to technologies that logically reason and predict by judging information, including knowledge / probability-based inference, optimization prediction, preference-based planning, and recommendation. Knowledge representation refers to technologies that automatically process human experiential information into knowledge data, including knowledge construction (data generation / classification) and knowledge management (data utilization). Motion control refers to technologies that control the autonomous driving of vehicles and the movement of robots, including motion control (navigation, collision, driving) and manipulation control (behavior control).

[0056] Generally, to apply machine learning algorithms to real-world situations, training is performed using a trial-and-error method due to the inherent characteristics of the fundamental methodologies. In particular, deep learning requires hundreds of thousands of iterations. Since it is impossible to execute this in a real physical external environment, training is instead performed through simulations that virtually recreate the actual physical environment on a computer.

[0057] In the present invention, Artificial Intelligence (AI) refers to a technology that imitates human learning ability, reasoning ability, and perceptual ability, and implements them on a computer, and may include concepts such as machine learning and symbolic logic. Machine Learning (ML) is an algorithmic technology that classifies or learns the characteristics of input data on its own. AI technology can analyze input data as a machine learning algorithm, learn from the results of the analysis, and make judgments or predictions based on the results of the learning. Furthermore, technologies that mimic the functions of the human brain, such as cognition and judgment, by utilizing machine learning algorithms can also be understood as falling within the category of AI. For example, technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control may be included.

[0058] Machine learning can refer to the process of training neural network models using experience in processing data. Through machine learning, computer software can improve its own data processing capabilities. A neural network model is constructed by modeling the correlations between data, and these correlations can be expressed by multiple parameters. A neural network model extracts and analyzes features from given data to derive correlations between them; machine learning can be defined as the process of optimizing the model's parameters by repeating this process. For example, a neural network model can learn the mapping (correlation) between inputs and outputs for data given as input-output pairs. Alternatively, even when only input data is provided, a neural network model can derive regularities between the given data and learn those relationships.

[0059] An artificial intelligence learning model or neural network model can be designed to implement the structure of the human brain on a computer and may include multiple network nodes that have weights and simulate neurons of a human neural network. The multiple network nodes may have interconnected relationships by simulating the synaptic activity of neurons, where neurons exchange signals through synapses. In an artificial intelligence learning model, multiple network nodes may be located in layers of different depths and exchange data according to convolutional connections. The artificial intelligence learning model may be, for example, an Artificial Neural Network (ANN) or a Convolutional Neural Network (CNN). As an embodiment, the artificial intelligence learning model may be machine learned according to methods such as supervised learning, unsupervised learning, and reinforcement learning. Machine learning algorithms for performing machine learning may include Decision Tree, Bayesian Network, Support Vector Machine, Artificial Neural Network, Ada-boost, Perceptron, Genetic Programming, and Clustering.

[0060] Among these, CNNs are a type of multilayer perceptron designed to use minimal preprocessing. CNNs consist of one or more convolutional layers and standard artificial neural network layers stacked on top, additionally utilizing weights and pooling layers. Thanks to this structure, CNNs can fully utilize two-dimensional input data. Compared to other deep learning architectures, CNNs demonstrate good performance in both image and audio fields. CNNs can also be trained using standard backpropagation. CNNs have the advantage of being easier to train than other feedforward artificial neural network techniques and using a small number of parameters.

[0061] Convolutional networks are neural networks comprising sets of nodes with bounded parameters. Many computer vision tasks have been significantly improved, driven by the increased size of available training data and the availability of computational power, combined with algorithmic advancements such as discriminative linear units and dropout training. In the case of massive datasets, such as those available for many tasks today, outfitting is not critical, and increasing the network size improves test accuracy. Optimal utilization of computing resources becomes a limiting factor. To address this, distributed, scalable implementations of deep neural networks can be employed.

[0062] The present invention relates to a method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data. More specifically, the invention relates to a method and system for automating the process flow from the drawing input stage to CAD design, CAM conversion, and CNC processing data generation when performing fiber laser processing and CNC processing on various metal plates such as steel plates, stainless steel plates, and aluminum plates, and to optimizing processing efficiency while minimizing the generation of burrs and dross by utilizing an artificial intelligence model learned based on metal processing process data.

[0063] In particular, the present invention relates to a method and system for automating a metal processing process to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned from metal processing process data, which performs intelligent process control technology that minimizes quality deviation and reduces post-processing by optimally controlling complex process variables such as laser output, processing speed, focal position, auxiliary gas conditions, material thickness, and material properties in real-time or based on prior prediction.

[0064] Generally, the processing of metal sheet materials such as steel plates, stainless steel, and aluminum is carried out by receiving 2D or 3D drawings from a customer, creating or modifying drawings using a CAD program, setting tool paths and processing conditions through CAM software, generating CNC code (G-code, etc.), and performing actual processing using a fiber laser cutter or CNC equipment.

[0065] Furthermore, when multiple parts are placed on a single sheet metal for cutting, a nesting process is performed beforehand to properly arrange the part shapes and maximize raw material utilization efficiency. The nesting process is a key step for minimizing the loss rate of the original sheet metal and reducing scrap generation.

[0066] However, the following problems exist in conventional technology.

[0067] First, nesting work relies heavily on the worker's experience and skill level.

[0068] When arranging multiple parts of different shapes and sizes on a single sheet of steel, significant experience is required to determine the optimal arrangement to maximize material yield, and even the results of automatic arrangement frequently require manual correction by an operator.

[0069] Second, the nesting results are not considered integrally with actual processing conditions.

[0070] Although factors such as spacing between parts, cutting sequence, and heat concentration areas affect machining quality, conventional systems operate by separating layout optimization and machining condition settings.

[0071] Third, the setting of processing conditions relies on the operator's experience.

[0072] Settings such as laser output, cutting speed, auxiliary gas pressure, focal position, feed rate, gas type (nitrogen, oxygen, air cut), nozzle position, and frequency vary depending on the material and thickness; if these settings are improper, various cutting quality defects occur due to aging or malfunction of components such as lenses, nozzles, and nozzle adapters mounted on the laser cutting header.

[0073] In particular, in conventional metal processing processes, cutting quality defects frequently occur, such as drilling failures or defects at the starting point during piercing, uneven cutting surfaces depending on the cutting direction, incomplete cutting at the edge, adhesion of moss-like structures, slag, or burrs to the processed surface, formation of dross beneath the cut, deterioration of head or processing precision (head misalignment, processing deviation, etc.), processing instability and thermal deformation caused by changes in cutting speed, and incomplete cutting after piercing.

[0074] These defects are caused by complex factors such as minute deviations in process conditions, changes in equipment status, differences in material properties, and heat accumulation effects, and have the limitation of being difficult to predict quantitatively in advance.

[0075] Fourth, when burrs and dross occur, additional deburring or post-processing is required, which reduces productivity and increases manufacturing costs.

[0076] Fifth, the CAD-CAM-CNC conversion process is fragmented, leading to data compatibility issues and repetitive modification work, and requires resetting the previous step when process conditions change.

[0077] Sixth, even under identical material and thickness conditions, quality variations occur depending on equipment status, nozzle wear, lens contamination, and ambient temperature; however, there is a lack of intelligent systems that systematically learn these factors and incorporate them into the process.

[0078] Consequently, conventional technology operates nesting optimization, processing condition settings, quality prediction, and equipment control independently of one another, relying mostly on post-processing correction or experience-based responses. As a result, problems such as increased raw material loss rates, the occurrence of burrs and dross, unstable cutting quality, increased post-processing costs, and reduced productivity continue to occur.

[0079] The present invention was devised to fundamentally resolve the problems of manual nesting dependency, experience dependency in setting process conditions, and poor cutting quality arising from the aforementioned conventional technology. It aims to provide a method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model trained on metal processing data capable of solving the following technical challenges.

[0080] First, the goal is to automate the nesting process of placing multiple parts onto a single metal plate and minimize raw material loss, while simultaneously providing an integrated nesting optimization structure that considers heat concentration and quality impacts resulting from placement positions and cutting sequences.

[0081] Second, by using an artificial intelligence model trained on metal processing data including cutting conditions, tool condition, material properties, equipment condition, and sensor data during processing, it is possible to predict in advance the possibility of cutting quality defects such as piercing defects, deterioration of cutting surface uniformity according to the cutting direction, edge failure, adhesion of slag and burrs to the processing surface, dross generation, deterioration of head precision, processing instability due to changes in cutting speed, and incomplete cutting after piercing, before or during processing.

[0082] Third, based on the above prediction results, an adaptive control structure is provided that automatically calculates multiple process conditions, including feed rate, spindle speed, cutting depth, laser output, focal position, auxiliary gas pressure, cutting path, and cutting sequence, and reflects them in real time to the equipment control unit, thereby suppressing the occurrence of the above defect in advance.

[0083] Fourth, it implements a self-improving process control system by feeding back process data collected in real-time during processing to an artificial intelligence model to correct prediction accuracy, and by reflecting even the quality inspection results after processing is completed as training data.

[0084] Fifth, the goal is to establish a data-driven prediction and control structure that differentiates itself from existing post-correction-centered metalworking methods by linking nesting optimization, defect prediction, automatic process condition adjustment, and equipment control into a single integrated platform.

[0085] A method and system for automating a metal processing process to minimize burr generation and optimize processing efficiency using an artificial intelligence model learned based on metal processing process data according to an embodiment of the present invention comprises: a memory; a communication unit; The system includes a processor connected to the communication unit and memory to execute program commands stored in the memory, wherein the processor collects metal processing process data in real time from a metal processing device, including cutting condition information, tool status information, material information, and sensor data during processing, and using the collected metal processing process data as input values, predicts the probability of burr occurrence during processing as a probability value between 0 and 1 through a pre-trained first model, and based on the probability of burr occurrence, recalculates the probability of burr occurrence by resetting the value obtained by changing at least one process variable among feed rate, spindle speed, cutting depth, and cutting path as the input value of the first model, and obtains process conditions by repeatedly performing the change of process variables and prediction by the first model until the recalculated probability of burr occurrence becomes less than a pre-set reference value, transmits the process conditions to the metal processing device to adjust the processing conditions in real time, and retrains the first model based on the metal processing process data additionally collected during the processing and the data obtained by applying the burr occurrence result received from the metal processing device or operator terminal after the completion of processing to the metal processing process data, and the first model includes feed rate, spindle speed, cutting It may be an artificial intelligence model trained to determine the possibility of burr generation during machining by using process variables including depth and cutting path, tool condition, material properties, and sensor data as input data, and burr generation results confirmed after machining is completed as ground truth data.

[0086] According to one embodiment of the present invention, the processor, upon receiving cutting condition information including two-dimensional or three-dimensional shape data for a metal part from the metal processing device or the operator terminal, executes a nesting algorithm based on the two-dimensional or three-dimensional shape data for the metal part to determine the shape, production quantity, material, thickness, cutting interval, thermal stress distribution by cutting direction, residual heat accumulation effect, laser heat affected zone (HAZ) or heat distribution generated by the tool, minimum gap between neighboring metal parts in a batch, batch angle, cutting path, and statistical or simulation-based residual scrap generation probability of the metal part; based on the nesting algorithm, obtains recommended process data including a batch drawing of the metal part, cutting sequence, and cutting path; sets the recommended process data as input variables of the first model to obtain a prediction result of the probability of burr generation through the first model; updates the process conditions calculated based on the prediction result of the probability of burr generation based on equipment status information including nozzle wear, lens contamination, tool wear, and temperature change collected in real time from the metal processing device; and the recommended process data and the Updated process information can be provided to the metal processing device in real time.

[0087] A processor according to an embodiment of the present invention stores work data collected for each worker from the metal processing device or the worker terminal, executes a plurality of second models that have learned worker-specific characteristics using the work data as input data, executes a third model learned to determine batch similarity categories using two-dimensional or three-dimensional shape data of a metal part as input data, calculates a worker priority based on a selected category according to the output result of the third model and previously stored information on burr occurrence and material loss rate for each category, selects a second-1 model for a worker of a certain rank or higher corresponding to the selected category among the plurality of second models based on the priority, performs an ensemble operation by applying a first weight corresponding to the priority to the output value of each of the selected plurality of second-1 models, derives work data in which the possibility of burr occurrence is below a preset threshold and the material loss rate is minimized, corrects the recommended process data based on the derived work data, and the work data includes the worker's drawing modification history, setting values, the type of metal processing device used, the type of metal to be worked on, the burr occurrence, and the material loss rate, and the setting values ​​are for the metal processing device At least one of the frequency of the output wave, processing speed, assist gas pressure, focus position, and nozzle gap is a value adjusted by the operator, and the second model is an artificial intelligence model trained to determine the possibility of burr generation and material loss rate when a specific operator performs work under specific process conditions, using the operator's work data and the burr generation result and material loss rate according to the work result as training data, and the third model may be an artificial intelligence model trained to determine a batch similar category corresponding to a target drawing using the shape and size of the workpiece, existing batch data, and process data as training data.

[0088] According to one embodiment of the present invention, the processor calculates a score for each worker with a rank of at least a certain level corresponding to the selected category based on the output value of each of the selected plurality of 2-1 models, sets a second weight for each worker for each of the selected plurality of 2-1 models based on the calculated score, and performs an ensemble operation by applying a first weight corresponding to the priority to the output value of each of the selected plurality of 2-1 models, wherein the ensemble operation is performed by additionally applying the second weight, and if a worker whose calculated score is below a preset threshold is identified, the 2-2 model corresponding to the identified worker among the selected plurality of 2-1 models is excluded from the selected plurality of 2-1 models and the 2-2 model is deleted, and if new work data is collected for the identified worker, the 2-3 model is created and retrained using the newly collected work data as training data for the identified worker, and the 2-3 model is formed using the work data of the worker and the burr generation result and material loss rate according to the work result as training data for the specific It may be an artificial intelligence model trained to determine the likelihood of burr generation and material loss rate when performing work under process conditions.

[0089] According to one embodiment of the present invention, the processor collects metal oxide data and vibration data in real time from a sensor device comprising at least one first sensor for detecting at least one metal oxide within a metal processing area of ​​the metal processing device and a second sensor for detecting vibration of a metal processing workbench; for each of the metal oxide data and the vibration data, performs a correction to remove the acquired data as noise data for a predetermined period of time from the start of processing of the metal processing device; determines an abnormal situation based on the degree of metal oxide generation and the degree of vibration generation based on the corrected data; if the result of the determination determines that a workpiece conforming to a predetermined defect criterion is generated, generates an alarm through the metal processing device or the operator terminal; and if a work stoppage command is received from the metal processing device or the operator terminal, provides a control command to the metal processing device to stop the operation of the metal processing device; stores the metal oxide data and the vibration data from the start of processing of the metal processing device to the stoppage of operation of the metal processing device as abnormal situation data; sets a work stoppage condition based on the degree of metal oxide generation and the level of vibration generation based on the accumulated abnormal situation data; and performs work in the work stoppage condition or the alarm generation situation. The amount of workpiece waste is obtained as work progress data from the above-mentioned operator terminal, and based on the above-mentioned work progress data, the average amount of workpiece waste is calculated according to the oxide generation concentration by metal type and the vibration frequency by metal type, and if the calculated average amount of workpiece waste exceeds a preset threshold ratio, a forced work stop value for the oxide generation concentration by metal type and the vibration frequency by metal type is set based on the above-mentioned work progress data to generate forced stop data, and the accumulated and stored forced work stop value,From a fourth model, which is an artificial intelligence model trained to determine work stoppage criteria and alarm response using the accumulated work progress data and corresponding waste amount data as training data, results are obtained regarding the determination of a forced stoppage situation and an alarm occurrence situation based on the metal oxide data and vibration data additionally collected during a future processing process; if a forced stoppage situation is determined during a future processing process based on the output of the fourth model, a control command to stop the operation of the metal processing device is provided to the metal processing device; and if an alarm occurrence situation is determined during a future processing process based on the output of the fourth model, an alarm can be generated through the metal processing device or the operator terminal.

[0090] A device according to one embodiment of the present invention may be combined with hardware and controlled by a computer program stored on a medium to execute any one of the operation methods by the processor of the system described above.

[0091] According to the method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model trained on metal processing process data of the present invention, the following effects can be expected.

[0092] First, by predicting the possibility of eight types of cutting quality defects occurring before or during processing and automatically adjusting process conditions, it becomes possible to implement prevention-oriented quality control rather than post-processing correction.

[0093] Second, piercing defects, incomplete cutting, burrs, and dross are significantly reduced, minimizing deburring and reprocessing steps, which in turn reduces production lead time and post-processing costs.

[0094] Third, processing instability caused by heat concentration and variations in cutting surface quality depending on the cutting direction is reduced, thereby improving cutting surface uniformity and precision.

[0095] Fourth, by integrating and optimizing nesting results and processing conditions, it is possible to achieve comprehensive quality stabilization that reflects the correlation between batch, cutting sequence, and quality, going beyond simple material yield improvement.

[0096] Fifth, through real-time data-based feedback control, it can adaptively respond to changes in equipment status (nozzle wear, optical system contamination, tool wear, etc.), thereby maintaining quality stability even during repetitive production under the same conditions.

[0097] Sixth, as process data and quality inspection results are continuously learned, the prediction accuracy of the artificial intelligence model gradually improves, enabling the implementation of a self-evolving manufacturing system that actively responds to changes in the production environment.

[0098] Seventh, automatic nesting and automatic setting of process conditions are possible even in multi-product, small-batch production environments, significantly improving production flexibility and the level of automation.

[0099] Consequently, the present invention functions not as a simple processing automation system, but as a data-based intelligent metal processing integrated control platform that predicts and controls cutting quality defects in advance, and can provide technical effects that simultaneously achieve burr reduction and productivity improvement.

[0100] Hereinafter, embodiments will be described in detail with reference to the attached drawings.

[0101] FIG. 1 is a block diagram schematically illustrating the basic configuration of a metal processing process automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data in one embodiment, FIG. 2 and FIG. 3 are diagrams schematically illustrating a metal processing process automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data according to one embodiment, and FIG. 4 and FIG. 5 are basic operation flowcharts of a metal processing process automation method for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data according to one embodiment.

[0102] Referring to FIG. 1, the system of the present invention may include an electronic device (100) comprising a processor (110), a memory (120), and a communication unit (130). In one embodiment, the electronic device (100) may be a server or a terminal. According to one embodiment, the processor (110) may be composed of one or more processors, configured to perform operations or data processing regarding the control and / or communication of each component of the electronic device (100). The memory (120) may store information related to the method described above or store a program in which the method described above is implemented. The memory (120) may be a volatile memory or a non-volatile memory.

[0103] According to one embodiment, the processor (110) can execute a program and control the electronic device (100). The code of the program executed by the processor (110) can be stored in memory (120). Operations of the processor (110) can be performed by loading instructions stored in memory (120). The electronic device (101) can be connected to an external device (e.g., a personal computer or a network) through an input / output device (not shown) and exchange data.

[0104] According to one embodiment, there are no limitations on the computation and data processing functions that the processor (110) can implement on the electronic device (100), but below, the process processing functions of the metal processing process automation platform for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data will be described.

[0105] Meanwhile, referring to FIGS. 2 and 3, the communication unit (130) can be connected to another device (e.g., metal processing device 10, worker terminal 20) through a data line (e.g., conductive line, wireless communication network, etc.). For example, the communication unit (130) can communicate with at least one of the metal processing device (10) and the worker terminal (20) to exchange signals and / or data.

[0106] Here, other devices may be servers or terminals such as electronic devices (100).

[0107] It goes without saying that other devices are not limited to FIGS. 2 and FIGS. 3 and can be provided in multiple numbers.

[0108] In addition, as shown in FIGS. 2 and 3, the metal processing device (10) and the worker terminal (20) do not correspond to the necessary and sufficient conditions of the method and system for automating metal processing processes to minimize burr generation and optimize processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data, and can be omitted or replaced.

[0109] Communication with the metal processing device (10) may be communication with a control unit including a processor and a communication unit provided in the metal processing device (10). For example, an electronic device (100) may be integrally provided in the metal processing device (10) and connected to an input device (e.g., touch panel, button, display, etc.) provided in the metal processing device (10), but is not limited thereto.

[0110] Meanwhile, referring to FIG. 4, the processor (110) can collect metal processing process data in real time (S410) from a metal processing device (10) and / or a worker terminal (20), including cutting condition information (e.g., 2D or 3D shape data for a metal part, feed rate, spindle rotation speed, cutting depth and cutting path, etc.), tool status information (e.g., laser type, type of processing tool, predicted lifespan based on the usage of the processing tool, etc.), material information (e.g., metal material to be processed, etc.) and sensor data during processing (e.g., vibration frequency generated during processing, laser heat affected zone (HAZ) detection data, heat distribution detection data by the processing tool, etc.).

[0111] For example, sensor data may be data detected through a vibration sensor, a heat sensing sensor, an optical sensor, a nozzle gas pressure sensor, etc. provided in a metal processing device (10), but is not limited thereto and may be modified or added. As an example, the vibration sensor may be configured as an acceleration sensor to measure an amplitude in the range of 0.01 g to 50 g in a frequency band of 1 Hz to 20 kHz, and the heat sensing sensor may be configured as an infrared pyrometer to measure a temperature in the range of 200°C to 2000°C within ±1% error. In addition, the nozzle gas pressure sensor may be configured to detect pressure in the range of 0.1 bar to 30 bar in units of 0.01 bar, and the gas flow sensor may be configured to measure a flow rate in the range of 1 L / min to 300 L / min. In a sensor device including the aforementioned sensors, the vibration sensor, the heat sensing sensor, and the nozzle gas pressure sensor are not limited to necessary and sufficient conditions and may be added, omitted, or replaced.

[0112] The processor (110) inputs the collected metal processing data into a pre-trained first model (S420) and obtains the probability p of burr occurrence during processing predicted as a probability value between 0 and 1 from the output of the first model (S430).

[0113] For example, the first model may be an artificial intelligence model trained to determine the possibility of burr generation during machining by using process variables including feed rate, spindle speed, cutting depth, and cutting path, tool condition, material properties, and sensor data as input data, and using the burr generation result confirmed after machining is completed as correct answer data.

[0114] At this time, if the probability of burr occurrence is greater than or equal to a preset reference value i, the processor (110) can recalculate the probability of burr occurrence by resetting (S440) the value of at least one process variable among feed rate, spindle rotation speed, cutting depth, and cutting path to the input value of the first model based on the probability of burr occurrence.

[0115] For example, the probability p of burr occurrence can be calculated by the following sigmoid function.

[0116] p = 1 / (1 + exp(-z)), where z = W·X + b, W is the weight vector, X is the input vector, and b is the bias value.

[0117] For example, the reference value i can be set in the range of 0.3 to 0.7.

[0118] By repeatedly performing changes to process variables and predictions by the first model until the recalculated probability of burr occurrence becomes less than a preset reference value, the processor (110) can obtain the input value of the first model, which makes the probability of burr occurrence less than a preset reference value, as a process condition (S450).

[0119] For example, changes to process variables can be performed by at least one of a grid search method that changes stepwise by set incremental values ​​within a set search range or an iterative optimization method based on gradient descent that minimizes a set objective function.

[0120] For example, the objective function can be defined as follows.

[0121] F = w1·material loss rate + w2·maximum thermal stress + w3·residual heat accumulation value, where w1, w2, and w3 are weights between 0 and 1 inclusive, satisfying w1+w2+w3=1.

[0122] For example, the first model may be a supervised learning-based artificial neural network model including a multi-layer perceptron or recurrent neural network structure, configured to have multiple hidden layers that repeatedly perform linear combination operations and non-linear activation functions on input vectors, and to calculate probability values ​​in the range of 0 to 1 by applying a sigmoid activation function in the final output layer, and may be a model trained to minimize a binary cross-entropy loss function using ground truth data as the confirmed occurrence of burrs after processing is completed. Additionally, retraining of the first model may be performed using an online additional training method that applies mini-batch stochastic gradient descent to new data while maintaining the existing training weights as initial values.

[0123] For example, a gradient descent-based iterative optimization method can be performed according to the following formula.

[0124] X_{t+1} = X_t - α∇L(X_t),

[0125] Here, α is a learning rate between 0.0001 and 0.1, and L is the binary cross-entropy loss function.

[0126] In this way, the processor (110) can transmit process conditions to the metal processing device (10) to adjust the processing conditions in real time (S460).

[0127] For example, the input vector of the first model may be composed of an n-dimensional real vector including feed rate (mm / s), spindle speed (rpm), cutting depth (mm), cutting path length (mm), material thickness (mm), laser power (W), gas pressure (bar), vibration RMS value (g), and average machining temperature (°C).

[0128] Meanwhile, referring to FIG. 5, after performing step S450, the processor (110) can retrain the first model (S520) based on the metal processing process data additionally collected during processing and the burr generation result received from the metal processing device (10) or the worker terminal (20) after processing is completed, which is applied to the metal processing process data (S510).

[0129] Meanwhile, a processor (110) according to one embodiment of the present invention may receive two-dimensional or three-dimensional shape data for a metal part from a metal processing device (10) or a worker terminal (20) while performing step S410.

[0130] The processor (110) can execute a nesting algorithm based on two-dimensional or three-dimensional shape data of the metal parts to determine the shape of the metal parts, production quantity, material, thickness, cutting interval, thermal stress distribution by cutting direction, residual heat accumulation effect, laser heat effect range or heat distribution generated by the tool, minimum gap between neighboring metal parts in the batch, batch angle, cutting path and statistical or simulation-based probability of residual scrap generation.

[0131] For example, a nesting algorithm may be an optimization algorithm that takes shape data and thermal stress distribution as input and arranges them to minimize a cost function.

[0132] For example, a nesting algorithm can quantify the thermal stress distribution and residual heat accumulation effects by cutting direction using a finite element analysis (FEM)-based thermal stress simulation or a regression model derived from past process data, and determine the placement location and cutting order so that the quantified thermal stress value is below a preset allowable threshold.

[0133] Accordingly, the nesting algorithm may be a mathematical optimization algorithm that includes a geometric collision detection operation for determining whether shape outlines collide and an objective function minimization operation for minimizing the material loss rate.

[0134] As the nesting algorithm is executed, the processor (110) can obtain recommended process data including a layout drawing of metal parts, a cutting sequence, and a cutting path based on the nesting algorithm.

[0135] In steps S420 and / or S440, the processor (110) can set recommended process data as input variables of the first model and obtain a result of predicting the likelihood of burr occurrence through the first model.

[0136] The processor (110) can update process conditions calculated based on a burr occurrence probability prediction result based on equipment status information collected in real time from the metal processing device (10), including nozzle wear of the laser cutter (e.g., change in gas pressure, change in gas flow rate, change in nozzle temperature, etc.), lens contamination of the laser cutter (e.g., lens temperature, laser transmittance, change in focal position, change in energy density, etc.), processing tool wear (e.g., usage, amount of vibration), and temperature change of the material and / or nozzle.

[0137] For example, equipment status information may be data collected based on sensor data.

[0138] For example, nozzle wear may be determined when (i) a pressure deviation of ±10% or more relative to the reference gas pressure occurs, (ii) a gas flow rate fluctuation rate of 5% or more, and (iii) the nozzle temperature rises by 15℃ or more relative to the reference temperature.

[0139] For example, lens contamination may be determined when (i) the laser transmittance decreases by more than 3% compared to the initial value, (ii) the protective lens temperature rises by more than 10°C compared to the reference temperature, or (iii) the focal position shows a deviation of more than ±0.1 mm.

[0140] Accordingly, in performing step S460, the processor (110) can provide recommended process data and updated process information to the metal processing device (10) in real time.

[0141] Meanwhile, a processor (110) according to one embodiment of the present invention can store work data collected by each worker from a metal processing device (10) or a worker terminal (20).

[0142] For example, work data may include the worker's drawing modification history, setting values, the type of metal processing device (10) used, the type of metal to be worked on, whether burrs are generated, and the material loss rate.

[0143] The set value may be a value adjusted by an operator for at least one of the frequency of the output wave for the metal processing device (10), the processing speed, the assist gas pressure, the focal position, and the nozzle gap.

[0144] The processor (110) can execute a plurality of second models that have learned the characteristics of each worker using work data as input data.

[0145] For example, the second model may be an artificial intelligence model trained to determine the probability of burr generation and material loss rate when a specific worker performs work under specific process conditions, using the worker's work data and the burr generation results and material loss rate based on the work results as training data.

[0146] Additionally, the processor (110) can execute a third model trained to determine a placement similarity category of a metal part, which is a workpiece, by using two-dimensional or three-dimensional shape data (shape and size of the workpiece) for the metal part as input data.

[0147] For example, a placement similar category may be a plurality of groups that match or have a similar degree of match according to the minimum gap between adjacent metal parts in the placement, the placement angle, etc., in the drawing in which the workpiece is placed on the material according to its size and shape.

[0148] The processor (110) can calculate the worker priority for the selected category based on the output result of the third model for the two-dimensional or three-dimensional shape data of the metal part, and the previously stored information on whether burrs occur and the material loss rate for each category.

[0149] The processor (110) can select a second-1 model for a worker of a certain rank or higher corresponding to a selected category among a plurality of second models based on priority for a selected category.

[0150] The processor (110) can perform an ensemble operation by applying a first weight corresponding to the priority to each output value of a plurality of selected 2-1 models to derive work data in which the possibility of burr generation is below a preset threshold and the material loss rate is minimized.

[0151] For example, the first weight can be calculated as a value normalized to be proportional to the inverse of the historical average burr occurrence rate and average material loss rate by category.

[0152] For example, the ensemble operation can be performed by multiplying the output probability value of each 2-1 model by a first weight and then performing a weighted sum.

[0153] The processor (110) can correct recommended process data based on the derived work data.

[0154] For example, the third model may be an artificial intelligence model trained to determine a layout similarity category corresponding to a target drawing using the shape and size of a workpiece, existing layout data, and process data as training data.

[0155] For example, the third model may be a classification model that generates a low-dimensional embedding vector through a pre-trained feature extraction network on the shape feature vector of an input workpiece, calculates a distance value between the embedding vector and a pre-stored category center vector, and classifies placement similar categories based on the minimum distance.

[0156] In one embodiment, a processor (110) according to one embodiment of the present invention can calculate a score for each worker of a certain rank or higher corresponding to a selected category based on the output value of each of a plurality of selected 2-1 models.

[0157] For example, the score per worker may be a quantitative score calculated by weighting the burr occurrence rate, average material loss rate, and work stability indicators over a recently set period.

[0158] For example, the score S per worker can be calculated as follows.

[0159] S = a·(1-burr rate) + b·(1-material loss rate) + c·work stability index, where a, b, and c are weights between 0 and 1 inclusive, and a+b+c=1.

[0160] The processor (110) can set a second worker-specific weight for each of the multiple selected second-1 models based on the calculated score.

[0161] For example, the second weight can be set by normalizing the score per worker to a range of 0 to 1.

[0162] In performing an ensemble operation by applying a first weight corresponding to the priority to each output value of a plurality of selected 2-1 models, the processor (110) may perform the ensemble operation by additionally applying a second weight for each worker.

[0163] At this time, when calculating a score for each worker of a certain rank or higher corresponding to a selected category, if a worker whose calculated score is below a preset threshold is identified, the processor (110) may exclude the 2-2 model corresponding to the identified worker from the selected plurality of 2-1 models and delete the 2-2 model.

[0164] For example, a worker whose calculated score is below a preset threshold may be determined to be a candidate for model deletion only if the score falls below the threshold for more than a preset number of consecutive times.

[0165] After deleting the 2-2 model, if new work data is collected for a worker matching the deleted 2-2 model, the processor (110) can create a 2-3 model and perform retraining using the newly collected work data as training data for the identified worker.

[0166] For example, the 2nd-3rd model may be an artificial intelligence model trained to determine the probability of burr generation and material loss rate when a worker performs work under specific process conditions, using the worker's work data and the burr generation results and material loss rate based on the work results as training data.

[0167] Accordingly, the retrained 2-3 model can be included in the 2nd model by replacing the deleted 2-2 model.

[0168] In one embodiment, a processor (110) according to one embodiment of the present invention can collect metal oxide data and vibration data in real time from a sensor device comprising at least one first sensor that detects at least one metal oxide within a metal processing area of ​​a metal processing device (10) and a second sensor that detects vibration of a metal processing workbench.

[0169] The processor (110) can perform correction to remove the acquired data as noise data for each of the metal oxide data and vibration data for a predetermined period of time from the start of processing of the metal processing device (10).

[0170] For example, noise removal correction can be performed by removing or smoothing data collected during a preset warm-up time interval after the start of processing using a moving average filter or a low-pass filter.

[0171] The processor (110) can determine an abnormal situation based on the corrected data, the degree of metal oxide generation and the degree of vibration generation.

[0172] For example, an abnormal situation can be defined as a case where the metal oxide concentration and vibration amplitude simultaneously exceed a preset first threshold and a second threshold, respectively.

[0173] For example, the first threshold value may be set in the range of a metal oxide concentration of 50 ppm or more and 500 ppm or less, and the second threshold value may be set in the range of a vibration RMS value of 1 g or more and 10 g or less. Accordingly, in determining an abnormal situation, if the metal oxide concentration and the vibration amplitude simultaneously exceed their respective threshold values ​​and this condition persists for 0.5 seconds or longer, it may be determined to be an abnormal situation.

[0174] If the processor (110) determines that the result of the judgment is a situation in which a workpiece conforming to a preset defect standard occurs, it can generate an alarm through the metal processing device (10) or the worker terminal (20).

[0175] For example, alarms can be output in various forms, such as sounds or pop-up messages.

[0176] When the processor (110) receives a work stop command from the metal processing device (10) or the worker terminal (20), it can provide a control command to the metal processing device (10) to stop the operation during the processing of a metal part of the metal processing device (10).

[0177] At this time, the processor (110) can store metal oxide data and vibration data as abnormal situation data from the time the metal processing device (10) starts processing until the time the metal processing device (10) stops operating upon receiving a work stoppage command.

[0178] Accordingly, the processor (110) can set operation suspension conditions based on the degree of metal oxide generation and the level of vibration generation, based on the accumulated stored abnormal situation data.

[0179] Meanwhile, the processor (110) can obtain the amount of waste material that has been processed under a work stoppage condition or an alarm occurrence condition as work progress data from the worker terminal (20).

[0180] The processor (110) can calculate the average amount of waste of workpieces based on the oxide generation concentration by metal type and the vibration frequency by metal type, based on work progress data.

[0181] When the average amount of waste of the calculated workpiece exceeds a preset threshold ratio, the processor (110) can generate forced stop data by setting a forced work stop value for the oxide generation concentration and vibration frequency of each metal type based on the work progress data.

[0182] For example, the forced work stoppage value may be a value calculated based on regression coefficients derived from the results of a multivariate regression analysis between the oxide concentration and frequency of each metal type and the average waste amount.

[0183] The processor (110) can obtain results of determining a forced stop situation and an alarm occurrence situation based on additional metal oxide data and vibration data collected during a future processing process from a fourth model, which is an artificial intelligence model trained to determine a work stop criteria and an alarm response using accumulated stored forced work stop values, accumulated stored work progress data, and corresponding waste amount data as training data.

[0184] For example, the fourth model may be composed of a supervised learning-based binary classification model that uses accumulated stored forced task interruption values ​​and task progress data as training data.

[0185] For example, the input data of the fourth model includes [oxide concentration (ppm), vibration RMS (g), processing time (sec), material type, laser output (W)], and the output data may be whether to forcibly stop (0 or 1).

[0186] If the processor (110) determines that a forced interruption situation is occurring during a future processing process based on the output of the fourth model, it can provide a control command to the metal processing device (10) to stop the operation of the metal processing device (10).

[0187] If the processor (110) determines that an alarm is likely to occur during a future processing process based on the output of the fourth model, it can generate an alarm through the metal processing device (10) or the worker terminal (20).

[0188] A device according to one embodiment of the present invention may be combined with hardware and controlled by a computer program stored on a medium to execute any one of the operation methods by the processor of the system described above.

[0189] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0190] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or they may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0191] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0192] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0193] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

Claims

Claim 1 In a metal processing automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model trained on metal processing process data, the system comprises: memory; a communication unit; The system includes a processor connected to the communication unit and memory and executing program commands stored in the memory, wherein the processor collects metal processing process data in real time from a metal processing device, including cutting condition information, tool status information, material information, and sensor data during processing, and using the collected metal processing process data as input values, predicts the probability of burr occurrence during processing as a probability value between 0 and 1 through a pre-trained first model, and based on the probability of burr occurrence, recalculates the probability of burr occurrence by resetting the value obtained by changing at least one process variable among feed rate, spindle speed, cutting depth, and cutting path as the input value of the first model, and obtains process conditions by repeatedly performing the change of process variables and prediction by the first model until the recalculated probability of burr occurrence becomes less than a pre-set reference value, transmits the process conditions to the metal processing device to adjust the processing conditions in real time, and retrains the first model based on the metal processing process data additionally collected during processing and the data obtained by applying the burr occurrence result received from the metal processing device or operator terminal after processing completion to the metal processing process data, wherein the first model includes feed rate, spindle speed, An artificial intelligence model trained to determine the possibility of burr generation during machining using process variables including cutting depth and cutting path, tool condition, material properties, and sensor data as input data, and burr generation results confirmed after machining completion as correct data, and the processor, upon receiving cutting condition information including 2D or 3D shape data for a metal part from the metal processing device or the operator terminal, based on the 2D or 3D shape data for the metal part, the shape of the metal part,Executing a nesting algorithm that determines the production quantity, material, thickness, cutting interval, thermal stress distribution by cutting direction, residual heat accumulation effect, laser heat-affected zone (HAZ) or heat distribution generated by the tool, minimum gap between adjacent metal parts in the batch, batch angle, cutting path, and statistical or simulation-based probability of residual scrap generation; based on the nesting algorithm, obtaining recommended process data including the batch drawing, cutting sequence, and cutting path of the metal parts; setting the recommended process data as input variables of the first model to obtain a prediction result of the probability of burr generation through the first model; updating the process conditions calculated based on the prediction result of the probability of burr generation based on equipment status information collected in real-time from the metal processing device, including nozzle wear, lens contamination, tool wear, and temperature change; providing the recommended process data and the updated process information to the metal processing device in real-time; storing work data collected by operator from the metal processing device or the operator terminal; executing a plurality of second models that have learned operator-specific characteristics using the work data as input data; and 2D or 3D shape data for the metal parts Execute a third model trained to determine batch similarity categories using input data, calculate a worker priority based on the selected category according to the output result of the third model and previously stored information on burr occurrence and material loss rate by category, select a 2-1 model for a worker of a certain rank or higher corresponding to the selected category among the plurality of 2 models based on the priority, and perform an ensemble operation by applying a first weight corresponding to the priority to the output value of each of the selected plurality of 2-1 models to derive work data in which the probability of burr occurrence is below a preset threshold and the material loss rate is minimized.A metal processing process automation system for minimizing burr generation and optimizing processing efficiency by utilizing an artificial intelligence model learned based on metal processing process data, characterized by: correcting the recommended process data based on the derived work data; wherein the work data includes the operator's drawing modification history, setting values, the type of metal processing device used, the type of metal to be processed, whether burrs occur, and the material loss rate; wherein the setting values ​​are values ​​adjusted by the operator for at least one of the frequency of the output wave for the metal processing device, processing speed, assist gas pressure, focus position, and nozzle gap; wherein the second model is an artificial intelligence model learned to determine the probability of burr generation and the material loss rate when a specific operator performs work under specific process conditions, using the operator's work data and the burr generation results and material loss rate according to the work results as learning data; and wherein the third model is an artificial intelligence model learned to determine a batch similar category corresponding to a target drawing, using the shape and size of the workpiece, existing batch data, and process data as learning data. Claim 2 delete Claim 3 delete

Images