KR102656853B1 - System for artificial intelligence-based cutting tool wear management and method thereof - Google Patents

Abstract

The present invention relates to an artificial intelligence-based cutting tool wear management system and method. More specifically, the artificial intelligence-based cutting tool wear management system according to the present invention includes a cutting machine equipped with a cutting tool and performing cutting processing; A sensor module installed in the cutting machine to measure the cutting tool movement value, spindle torque value, etc., a terminal module connected to the sensor module and provided to input and output the measured data, and measurement from the sensor module. An embedded computer that receives data and predicts machining defects according to the life of the cutting tool through an artificial intelligence analysis algorithm to determine whether the cutting tool is worn, and generates an alarm if the embedded computer determines that the cutting tool is worn. Includes an alarm module that

Description

Artificial intelligence-based cutting tool wear management system and method thereof {System for artificial intelligence-based cutting tool wear management and method thereof}

The present invention relates to an artificial intelligence-based cutting tool wear management system and method. More specifically, artificial intelligence is capable of monitoring the status of cutting tools to accurately identify wear progress and prevent damage so that cutting tools can be replaced at an appropriate time. It relates to a basic cutting tool wear management system and method.

Cutting processing is a method of cutting and processing materials, and such cutting processing inevitably uses cutting tools.

In a production system using machine tools, wear and damage to cutting tools are inevitable due to friction on the contact surface of the machined parts, which results in inaccuracy, reduced reliability, wasted time due to cutting tool exchange time, and requires machine stoppage. Problems such as frequent replacement occurred.

In the production system, cutting tools mostly rely on visual inspection methods and regular replacement according to usage time, and require automation.

Wear and damage to cutting tools resulting from the limitations of traditional visual inspection methods can cause serious problems in not only work safety and loss of processing time, but also performance and product quality deterioration, so cutting tools cannot be replaced at an appropriate time. A monitoring system is needed to accurately identify wear progress and prevent damage.

Republic of Korea Patent No. 10-0979032 (2010.08.24.)

The purpose of the present invention is

An artificial intelligence-based cutting tool wear management system that collects real-time data on cutting tool use and uses this to analyze the wear status of cutting tools to prevent immediate machine tool damage and mass defects in the event of a cutting tool abnormality. That method is provided.

The objects of the present invention are not limited to the objects mentioned above, and other objects not mentioned can be clearly understood by those skilled in the art from the description below.

In order to achieve the above purpose, the artificial intelligence-based cutting tool wear management system according to the present invention includes a cutting machine equipped with a cutting tool to perform cutting, a cutting tool installed on the cutting machine, a movement value of the cutting tool, and spindle torque. A sensor module that measures values, etc., a terminal module that is connected to the sensor module and is equipped to input and output the measured data, and receives the measured data from the sensor module to determine the life of the cutting tool through an artificial intelligence analysis algorithm. It includes an embedded computer that predicts machining defects and determines whether the cutting tool is worn, and an alarm module that generates an alarm when the embedded computer determines that the cutting tool is worn.

In addition, the embedded computer includes a data collection unit that collects measurement data including cutting tool movement values and spindle torque values from the sensor module, and processes the measurement data into meaningful values for application to the artificial intelligence analysis algorithm. A data pre-processing unit, a data learning unit that receives the processed data from the data pre-processing unit and learns a prediction model to predict machining defects according to the life of the cutting tool, and artificial intelligence analysis after completion of learning in the data learning unit. It is characterized by including a data analysis unit that determines whether the cutting tool is normal or worn through an algorithm.

In addition, the embedded computer is characterized by including a monitoring unit that visually provides measurement data collected from the sensor module through signal processing so that the user can check work progress information in real time.

In addition, the artificial intelligence analysis algorithm is characterized as an XGBoost (eXtreme Gradient Boosting, XGB) algorithm that creates a strong prediction model by weighting the learning errors of weak prediction models and sequentially reflecting them in the next prediction model.

Meanwhile, the artificial intelligence-based cutting tool wear management method according to another aspect of the present invention is an artificial intelligence-based cutting tool wear management method performed by a computer, in which the cutting tool movement value and spindle torque value are collected from the sensor module in the data collection unit. Collecting measurement data including, processing the data into meaningful values for applying the measurement data collected in the data pre-processing unit to an artificial intelligence analysis algorithm, data processed from the data pre-processing unit in the data learning unit. A step of receiving input and learning a prediction model to predict machining defects according to the life of the cutting tool. After completing the learning in the data learning unit, the data analysis unit determines whether the cutting tool is normal or worn through an artificial intelligence analysis algorithm. In the step of determining whether the cutting tool is worn, the alarm module generates a warning alarm when it is determined that the cutting tool is worn.

In addition, the step of processing the data includes receiving raw data measured in real time from a sensor module, first processing the raw data to reduce meaningless values, omissions, and errors based on the raw data, It includes the step of classifying the data values that have been processed through the primary processing step, identifying the characteristics to learn the prediction model, and performing secondary processing.

In addition, the artificial intelligence analysis algorithm is characterized as an XGBoost (eXtreme Gradient Boosting, XGB) algorithm that creates a strong prediction model by weighting the learning errors of weak prediction models and sequentially reflecting them in the next prediction model.

The artificial intelligence-based cutting tool wear management system and method according to the present invention prevents damage to machine tools by allowing managers or workers to check through an immediate alarm when a cutting tool abnormality occurs, thereby inducing quick action against the abnormality. And the occurrence of mass defects can be prevented.

In addition, by providing an accurate condition for cutting tool wear, there is an effect of reducing costs due to use of the tool life in an optimal state, improving the company's productivity, and securing the company's competitiveness in terms of quality.

Figure 1 is a configuration diagram showing an artificial intelligence-based cutting tool wear management system according to the present invention.
Figure 2 is a configuration diagram showing an embedded computer constituting an artificial intelligence-based cutting tool wear management system according to the present invention.
Figure 3 is a flowchart showing the overall flow of the artificial intelligence-based cutting tool wear management method according to the present invention.
Figure 4 is an example diagram illustrating signal characteristics by type for determining the wear state of a cutting tool according to the artificial intelligence-based cutting tool wear management method according to the present invention.

The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and will be implemented in various different forms. The present embodiments only serve to ensure that the disclosure of the present invention is complete and that common knowledge in the technical field to which the present invention pertains is not limited. It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims.

Specific details for implementing the present invention will be described in detail with reference to the accompanying drawings below. Regardless of the drawings, the same reference numerals refer to the same elements, and “and/or” includes each and all combinations of one or more of the mentioned items.

The terminology used herein is for describing embodiments and is not intended to limit the invention. As used herein, singular forms also include plural forms, unless specifically stated otherwise in the context. As used in the specification, “comprises” and/or “comprising” does not exclude the presence or addition of one or more other elements in addition to the mentioned elements.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used with meanings that can be commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless clearly specifically defined.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

Figure 1 is a configuration diagram showing an artificial intelligence-based cutting tool wear management system according to the present invention.

Referring to Figure 1, the artificial intelligence-based cutting tool wear management system largely includes a cutting machine 100, a sensor module 200, a terminal module 300, an embedded computer 400, and an alarm module 500. .

First, the cutting machine 100 refers to a machine tool that performs cutting using a cutting tool, and may generally mean a CNC (Computerized Numerical Control) lathe or a machining center (MCT), and is limited thereto. That is not the case.

Next, the sensor module 200 is a sensor that measures the actual cutting tool movement value compared to the setting value provided by the cutting machine 100, current and voltage data of the cutting machine 100, and spindle torque value, etc. It may consist of, but is not limited to. Here, of course, each sensor can be installed in a separate location on the cutting machine 100 to measure each data or value.

Next, the terminal module 300 is connected to the sensor module 200 and is a device for transmitting or receiving output of measured data and is equipped to input and output measured data.

As shown in FIG. 1, the terminal module 300 may be provided with a terminal unit 300a for outputting data and a master unit 300b for receiving output data, and the terminal unit 300a Of course, there may be more than one sensor depending on the location where each sensor is provided.

Next, the embedded computer 400 refers to a computer that receives the measured data from the sensor module 200 and determines whether the cutting tool is worn by predicting machining defects according to the life of the cutting tool through an artificial intelligence analysis algorithm. .

Here, the embedded computer 400 may refer to an industrial embedded computer with a built-in microprocessor capable of implementing computer functions, but is not limited thereto and may replace a general personal computer (PC), etc. Of course.

Figure 2 is a configuration diagram showing an embedded computer constituting an artificial intelligence-based cutting tool wear management system according to the present invention.

Referring to FIG. 2, to describe the embedded computer 400 in more detail, the embedded computer 400 includes a data collection unit 410, a data preprocessing unit 420, a data learning unit 430, and a data It may include an analysis unit 440.

First, the data collection unit 410 serves to collect the measurement data including the cutting tool movement value, spindle torque value, etc. from the sensor module 200.

Preferably, the data collection unit 410 receives measurement data output from the terminal unit 300a through the master unit 300b provided on the embedded computer 400 and collects the data.

Next, the data pre-processing unit 420 processes the data into meaningful values for application to the artificial intelligence analysis algorithm. Meaningful values here mean that measurements that occur in actual processes may have errors, noise values, abnormal data, etc., and these data can be interpreted as meaningless values, and values excluding these values can be interpreted as meaningless values. It can be interpreted as a meaningful value.

These meaningless, missing, or errors degrade the quality (or accuracy) of the data. Performing preprocessing with the above meaningful values to improve the quality in data analysis is one of the most important steps in data analysis. It can mean.

Additionally, the data pre-processing unit 420 may be configured to undergo primary and secondary pre-processing of the measurement data collected from the sensor module 200, that is, raw data.

The first preprocessing process may refer to a process of identifying the index of raw data and improving the quality index of low-quality data that deteriorates quality due to the collected raw data.

The secondary preprocessing process may refer to a process of converting the primary processed data processed through the primary preprocessing process into data for learning a prediction model to be described later, and attribute values required for learning the prediction model. This can mean the process of filtering, discarding unnecessary data, and leaving only data for actual use.

Next, the data learning unit 430 receives the processed data from the data pre-processing unit 420 and serves to learn a prediction model to predict machining defects according to the life of the cutting tool.

Next, the data analysis unit 440 serves to determine whether the cutting tool is normal or worn through an artificial intelligence analysis algorithm after the data learning unit 430 completes learning.

The artificial intelligence analysis algorithm classifies and defines signal patterns for cutting tool movement values (e.g., X, Y, Z axes) and spindle torque values collected from the sensor module 200 by type to determine cutting tool wear. , refers to an algorithm that analyzes the wear level of cutting tools by diagnosing the wear condition by comparing the current signal with the normal signal. This can be understood as determining a wear state when the current wear value of the cutting tool is equal to or higher than a preset reference value.

Here, the artificial intelligence analysis algorithm may refer to the XGBoost (eXtreme Gradient Boosting, Of course, this may also apply to the prediction model described in the data learning unit 430.

In determining cutting tool wear, the XGBoost algorithm can have excellent prediction performance in the classification and regression areas, overfitting regulation, and the ability to reduce the number of divisions by pruning divisions without positive gain.

In addition, cross-validation is performed internally every time it is repeated, so it has an optimized number of repetitions. If the evaluation value of the evaluation data set is optimized through cross-validation rather than the specified number of repetitions, the repetition can be stopped in the middle, thereby improving the processing process. There is an advantage to reducing it.

Meanwhile, the embedded computer 400 may further include a monitoring unit 450 that visually provides measurement data collected from the sensor module through signal processing so that the user can check work progress information in real time. .

The monitoring unit 450 may be in the form of a commonly configured touch display, and detailed description will be omitted.

Next, the alarm module 500 generates an alarm when the embedded computer 400 determines that the cutting tool is in a worn state, and may mean an industrial warning light.

The alarm module 500 is connected to the embedded computer 400 and is provided to effectively prepare for work stoppage by setting the entire warning range for complete wear of the cutting tool. The alarm module 500 provides a warning or alarm. This has the effect of allowing managers or workers to quickly respond to wear and tear of cutting tools.

In the following, an artificial intelligence-based cutting tool wear management method using the artificial intelligence-based cutting tool wear management system according to the present invention will be described.

Figure 3 is a flowchart showing the overall flow of the artificial intelligence-based cutting tool wear management method according to the present invention.

Referring to FIG. 3, to describe the artificial intelligence-based cutting tool wear management method according to the present invention, in the artificial intelligence-based cutting tool wear management method performed by a computer (or embedded computer), first, a data collection unit ( In step 410, measurement data including cutting tool movement values, spindle torque values, etc. are collected from the sensor module 200 (S110).

Next, a step is performed to process the measurement data collected by the data pre-processing unit 420 into meaningful values for application to the artificial intelligence analysis algorithm (S120).

The data processing step (S120) includes receiving raw data measured in real time from the sensor module 200, and first processing the raw data to reduce meaningless values, omissions, and errors based on the raw data. It may be configured to perform a step of classifying the data values primarily processed through the first processing step, identifying characteristics to learn the prediction model, and performing a second processing step.

The first processing step determines the index of the raw data and performs a process of improving the quality index for low-quality data that degrades the quality due to the collected raw data.

Thereafter, the secondary processing step is a step of converting the primary processing data processed through the first processing step into data for training a prediction model to be described later, filtering the attribute values required for learning the prediction model, and A process is performed to discard unnecessary data and leave only data for actual use.

Next, the data learning unit 430 receives the processed data from the data pre-processing unit 420 and performs a step of learning a prediction model to predict machining defects according to the life of the cutting tool (S130).

Next, after completing learning in the data learning unit, the data analysis unit performs a step of determining whether the cutting tool is normal or worn through an artificial intelligence analysis algorithm (S140).

To explain this as an example, the artificial intelligence analysis algorithm uses signal patterns for cutting tool movement values (e.g., X, Y, Z axes) and spindle torque values collected from the sensor module 200 to determine cutting tool wear. By classifying and defining each type, you can analyze the wear level of the cutting tool by comparing the current signal and the normal signal to diagnose the wear condition.

Figure 4 is an example diagram illustrating signal characteristics by type for determining the wear state of a cutting tool according to the artificial intelligence-based cutting tool wear management method according to the present invention.

To further explain with reference to FIG. 4, (a) of FIG. 4 shows a signal pattern indicating the initial setting state, and wear limit lines (red lines) 1 and 2 indicate wear limit lines. In addition, the damage limit lines (green lines) 4 and 5 are limit lines that designate the normal range of the cutting tool, and if the cutting tool is normal, a signal is located between the wear limit line and the damage limit line.

Figure 4(b) shows the signal pattern for the wear state of the cutting tool, and when the wear limit line is reached, it is judged to be in the worn state. In other words, it can be understood that if the current wear value of the cutting tool is equal to or higher than the preset standard value, it is judged to be in a wear state.

Figure 4 (c) shows the signal pattern for the damage state of the cutting tool, which may mean an abnormal pattern signal. If it does not reach (exceed) the range of the normal damage limit line, the cutting tool can be judged to be in a damaged state. .

In this way, the artificial intelligence analysis algorithm that determines the wear state of the cutting tool is the XGBoost (eXtreme Gradient Boosting, can be used, and the XGBoost (eXtreme Gradient Boosting,

In the step of determining wear, if it is determined that the cutting tool is worn, the alarm module 500 performs a step of generating an alarm (S150).

The alarm module 500 can generate a work stop alarm and a warning alarm, and the alarm generating step (S150) sets a warning range before the work stop alarm to efficiently prepare for work stoppage. It can be configured to generate. Here, the work stop alarm refers to an alarm for a situation in which equipment (work) must be stopped, and the warning alarm warns of a point in time when machining precision decreases due to wear, and the manager (operator) sends a warning alarm. It can lead to a quick response to wear of cutting tools.

Meanwhile, if the determination result in the step of determining whether the cutting tool is normal or worn (S140) is that the cutting tool is normal, the step of collecting the measurement data (S110) is repeated.

Therefore, the artificial intelligence-based cutting tool wear management system and method according to the present invention are designed to monitor the wear progress status (normal, replacement required) of the cutting tool in real time, so that the cutting tool can be replaced at an appropriate time. It is effective in preventing serious problems in performance and quality deterioration of products as well as loss of safety and processing time.

Although embodiments of the present invention have been described with reference to the above and the attached drawings, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing the technical idea or essential features. You will understand that it exists. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive.

100: Cutting machine
200: Sensor module
300: Terminal module
400: Embedded computer
410: Data collection unit
420: Data preprocessing unit
430: Data learning unit
440: Data analysis department
450: Monitoring unit
500: Alarm module

Claims (7)

delete delete delete delete In an artificial intelligence-based cutting tool wear management method performed by a computer,
Collecting measurement data including cutting tool movement value, spindle torque value, etc. from the sensor module in the data collection unit;
Processing the measurement data collected by the data pre-processing unit into meaningful values for application to an artificial intelligence analysis algorithm;
A data learning unit receiving the processed data from the data pre-processing unit and learning a prediction model to predict machining defects according to the life of the cutting tool;
After completing learning in the data learning unit, the data analysis unit determines whether the cutting tool is normal or worn through an artificial intelligence analysis algorithm; and
In the step of determining whether the cutting tool is worn, the alarm module generates a warning alarm when it is determined that the cutting tool is worn.
The step of processing the data is,
A step of receiving raw data measured in real time from a sensor module,
In order to reduce meaningless values, omissions, and errors based on the raw data, the quality index for the raw data is identified, and the quality index is improved for low-quality data that reduces quality due to the collected raw data. A first processing step,
Through the primary processing step, the primary processed data values are classified and the attribute values required to learn the prediction model are filtered to identify the characteristics, unnecessary data is discarded, and only data for actual use is left. It includes a data secondary processing step, which is the process of converting it into data for learning a prediction model,
The artificial intelligence analysis algorithm provides excellent prediction performance in the classification and regression areas in determining cutting tool wear, regulates overfitting, reduces the number of divisions by pruning divisions without positive gain, and internally crosses each iteration. It is configured to perform verification and has an optimized number of iterations. If the evaluation value of the evaluation data set is optimized through cross-validation rather than the specified number of iterations, the iteration can be stopped in the middle to reduce the processing process of weak prediction models. An artificial intelligence-based cutting tool wear management method characterized by the XGBoost (eXtreme Gradient Boosting, XGB) algorithm that creates a strong prediction model by weighting learning errors and sequentially reflecting them in the next prediction model. delete delete