KR20230072741A - System and method for diagnosing abnormal operation of machine tool based on deep learning - Google Patents

Abstract

The present specification relates to a technology for diagnosing an abnormal motion of a machine tool, and a system for diagnosing an abnormal motion of a machine tool based on deep learning according to an embodiment is installed on a table of a machine tool and measured by an accelerometer and an accelerometer that measure acceleration when the table moves. A STFT processing unit that generates an STFT image by performing STFT (Short Time Fourier Transform) processing on the accelerated acceleration data, and a deep learning model that uses the STFT image from the STFT processing unit to create a deep learning model for determining abnormal operation of a machine tool. It may include a learning model generation unit and an abnormal operation determining unit that calls the deep learning model, inputs the STFT image from the STFT processing unit to the deep learning model, and determines an abnormal operation type of the corresponding acceleration data.

Description

Deep learning-based machine tool abnormal operation diagnosis system and method {SYSTEM AND METHOD FOR DIAGNOSING ABNORMAL OPERATION OF MACHINE TOOL BASED ON DEEP LEARNING}

The present invention relates to a technology for diagnosing abnormal operation of a machine tool, and more particularly, to a system and method for diagnosing abnormal operation of a machine tool based on deep learning capable of diagnosing abnormal operation occurring in various parts of a machine tool based on deep learning.

In general, a machine tool refers to a machine used for the purpose of processing a metal/non-metal workpiece into a desired shape and dimension using an appropriate tool by various cutting or non-cutting processing methods.

Various types of machine tools are known according to the processing range, and a representative example is a CNC (Computerized Numerical Control) machine tool. A CNC machine tool automatically determines the position of a tool and processes a workpiece by directly connecting a program that records numerical data that selectively instructs the machining dimensions, shape, required tools, feed speed, etc. of a workpiece with a computer.

Depending on the use of machine tools, wear or damage of tools, wear or damage of bearings, and spindle runout abnormalities frequently occur. Currently, chatter that occurs during machining is identified through the acceleration response generated from the spindle. In addition, there is no diagnostic technology for identifying failures in other parts, so it is possible to diagnose only the abnormal operation of a very limited part.

Since it is not possible to diagnose the overall abnormal operation of the machine tool by identifying the chatter that occurs in the machine tool, proper maintenance is not performed, resulting in frequent breakdowns of the machine tool. There are problems with exposure to accidents.

Therefore, there is a need for a technique capable of diagnosing abnormal operations occurring in various parts of a machine tool.

This background art is technical information that the inventor possessed for derivation of the present invention or acquired in the process of derivation of the present invention, and cannot be said to be known art disclosed to general notaries prior to filing of the present invention.

Embodiments disclosed in this specification to solve the problems of the prior art have a technical task of providing a system and method for diagnosing abnormal operation of a machine tool based on deep learning capable of diagnosing an abnormal operation occurring in a machine tool.

In addition, the technical task of the embodiments is to provide a system and method for diagnosing abnormal motions of machine tools based on deep learning capable of diagnosing various types of abnormal motions occurring in machine tools.

In addition, embodiments make it a technical task to provide a system and method for diagnosing abnormal motions of a machine tool based on deep learning capable of diagnosing abnormal motions occurring in various parts of a machine tool.

The technical problem of the present invention is not limited to the above-mentioned matters, and from the following description, those of ordinary skill in the art to which the present invention belongs will be able to clearly understand other problems intended by the present invention. .

As a technical means for achieving the above-described technical problem, the system for diagnosing abnormal motion of a deep learning-based machine tool according to an embodiment of the present invention is installed on a table of a machine tool and includes an accelerometer and an accelerometer that measure acceleration when the table moves. A STFT processing unit that generates an STFT image by performing STFT (Short Time Fourier Transform) processing on the acceleration data measured by the STFT processing unit. Using the STFT image from the STFT processing unit, a deep learning model is created to determine abnormal operation of the machine tool. It may include a deep learning model generation unit that generates a deep learning model, and an abnormal operation determination unit that calls the deep learning model and inputs the STFT image from the STFT processing unit to the deep learning model to determine an abnormal operation type of the corresponding acceleration data.

According to an embodiment of the present invention, the STFT processing unit STFT-processes the learning acceleration data used to generate the deep learning model, and provides the learning STFT image generated by the deep learning model generation unit to the deep learning model generator, and determines abnormal operation of the machine tool. A target STFT image generated by STFT-processing the target acceleration data used to perform the target acceleration may be provided to the abnormal operation determining unit.

According to an embodiment of the present invention, the deep learning model may be an autoencoder-based model.

According to an embodiment of the present invention, the abnormal motion determiner may determine the abnormal motion type classified with the highest number of times as the abnormal motion type of the corresponding acceleration data among the abnormal motion types classified through the deep learning model.

Specific details according to various examples of the present invention other than the means for solving the problems mentioned above are included in the description and drawings below.

According to the embodiments described herein, a system and method for diagnosing an abnormal operation of a machine tool based on deep learning capable of diagnosing an abnormal operation occurring in a machine tool may be provided.

Further, according to the embodiments, a system and method for diagnosing various kinds of abnormal operations of a machine tool based on deep learning capable of diagnosing various kinds of abnormal operations occurring in a machine tool may be provided.

Further, according to embodiments, a system and method for diagnosing abnormal operations of a machine tool based on deep learning capable of diagnosing abnormal operations occurring in various parts of a machine tool may be provided.

By diagnosing the abnormal operation of the machine tool using this deep learning-based machine tool abnormal operation diagnosis technology, the overall abnormal operation of the machine tool can be diagnosed, and maintenance can be performed through diagnosis before a machine tool failure occurs. Since this can be done, it is possible to prevent the failure of the machine tool, to extend the service life of the machine tool, and to prevent workers from being exposed to accidents.

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

Since the contents of the problem to be solved, the means for solving the problem, and the effect mentioned above do not specify essential features of the claims, the scope of the claims is not limited by the matters described in the contents of the invention.

The accompanying drawings are provided to aid understanding of the embodiments of the present invention, and provide examples along with detailed descriptions. However, the technical features of this embodiment are not limited to specific drawings, and features posted in each drawing may be combined with each other to form a new embodiment.
1 is a diagram showing the configuration of a system for diagnosing an abnormal operation of a machine tool based on deep learning according to an embodiment of the present invention.
2 is a diagram for explaining the result of STFT processing of acceleration data in the system for diagnosing abnormal motion of a machine tool based on deep learning according to an embodiment of the present invention.
3 is a flowchart illustrating a method for diagnosing an abnormal operation of a machine tool based on deep learning according to an embodiment of the present invention.

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

The shapes, sizes, ratios, angles, numbers, etc. disclosed in the drawings to describe the embodiments of the present invention are illustrative, so the present invention is not limited to the details shown. Like reference numbers designate like elements throughout the specification. In addition, in describing the present invention, if it is determined that a detailed description of related known technologies may unnecessarily obscure the subject matter of the present invention, the detailed description will be omitted. When "includes", "has", "consists of", etc. mentioned in this specification is used, other parts may be added unless "only" is used. In the case where a component is expressed in the singular, the case including the plural is included unless otherwise explicitly stated.

In interpreting the components, even if there is no separate explicit description of the error range, it is interpreted as including the error range.

In the case of a description of a positional relationship, for example, when the positional relationship of two parts is described as "on", "upper", "at the bottom", "next to", etc., for example, "right" Or, unless "directly" is used, one or more other parts may be located between the two parts.

In the case of a description of a temporal relationship, when a temporal precedence relationship is described with “after,” “next to,” “next to,” “before,” etc., when it is not continuous unless “immediately” or “directly” is used may also be included.

Although first, second, etc. are used to describe various components, these components are not limited by these terms. These terms are only used to distinguish one component from another. Therefore, the first component mentioned below may also be the second component within the technical spirit of the present invention.

In describing the components of the present invention, terms such as first, second, A, B, (a), and (b) may be used. These terms are only used to distinguish the component from other components, and the nature, sequence, order, or number of the corresponding component is not limited by the term. When an element is described as being “connected”, “coupled” or “connected” to another element, that element is directly connected or may be connected to the other element, but indirectly unless specifically stated otherwise. It should be understood that other components may be "interposed" between each component that is or can be connected.

“At least one” should be understood to include all combinations of one or more of the associated elements. For example, "at least one of the first, second, and third elements" means not only the first, second, or third elements, but also two of the first, second, and third elements. It can be said to include a combination of all components of one or more.

Each feature of the various embodiments of the present specification can be partially or entirely combined or combined with each other, technically various interlocking and driving are possible, and each embodiment can be implemented independently of each other or can be implemented together in an association relationship. may be

Hereinafter, looking at the embodiments of the present invention through the accompanying drawings and embodiments are as follows. Since the scales of the components shown in the drawings have different scales from actual ones for convenience of explanation, they are not limited to the scales shown in the drawings.

Hereinafter, a system and method for diagnosing an abnormal operation of a machine tool based on deep learning according to an embodiment of the present invention will be described with reference to the accompanying drawings.

1 is a diagram showing the configuration of a system for diagnosing an abnormal operation of a machine tool based on deep learning according to an embodiment of the present invention.

Referring to FIG. 1 , a system for diagnosing an abnormal operation of a deep learning-based machine tool (hereinafter referred to as a diagnostic system) 1 according to an embodiment of the present invention includes an accelerometer 100, an STFT processing unit 200, and a deep learning model generating unit 300. ), and an abnormal operation determining unit 400, and the configuration of the diagnosis system 1 is not limited thereto.

The accelerometer 100 is installed on the machine tool C, is installed on a table T that moves according to the driving of the motor M, measures acceleration when the table T moves, and stores the measured acceleration data. It can be provided to the STFT processing unit 200. For example, one accelerometer 100 may be installed on the table T and may be installed in multiple areas of the table T.

For example, the accelerometer 100 may be installed in an edge region including a corner of the table T, a central region, or the like, and may be installed on an upper surface, a lower surface, or a side surface of the table T.

The STFT processing unit 200 may generate a STFT image by performing Short Time Fourier Transform (STFT) processing on acceleration data from the accelerometer 100 .

2 is a diagram for explaining the result of STFT processing of acceleration data in a system for diagnosing abnormal motion of a machine tool based on deep learning according to an embodiment of the present invention. In FIG. 2, (a) is acceleration data input to the STFT processing unit. , (b) is an STFT image generated by STFT processing by the STFT processor.

As shown in FIG. 2 , the STFT processing unit 200 may generate an STFT image having information on both time and frequency through STFT processing on acceleration data.

The STFT processing unit 200 may generate an STFT image to conform to a predetermined standard, and may convert the STFT image into a square image of a predetermined size, for example.

Acceleration data used to generate a deep learning model is referred to as learning acceleration data, and acceleration data used to determine an abnormal operation of a machine tool using a deep learning model is referred to as target acceleration data.

An STFT image generated based on learning acceleration data is referred to as a training STFT image, and an STFT image generated based on target acceleration data is referred to as a target STFT image.

The STFT processor 200 may provide the training STFT image to the deep learning model generator 300 and the target STFT image to the abnormal operation determiner 400 .

Since the training STFT image generated by the STFT processing unit 200 is provided to the deep learning model generation unit 300 that generates a deep learning model, the STFT processing unit 200 generates and stores the training STFT image, deep learning In order to facilitate processing in the model, the training STFT image may be converted into a square gray scale image of a predetermined size, stored, and provided to the deep learning model generator 300.

The deep learning model generation unit 300 inputs the learning STFT image from the STFT processing unit 200 into a script for generating a deep learning model for determining an abnormal operation (or failure) of a machine tool to achieve a certain level of accuracy. It is possible to create a deep learning model for determining the abnormal operation of the machine tool by changing the hyper parameters until

The deep learning model generator 300 may store the generated deep learning model in a database (DB) 500 .

Since it is difficult to acquire abnormal data in the case of machine tools, the deep learning model may be an autoencoder-based model, which is a model capable of learning even when normal/abnormal data is imbalanced, unlike general classification machine learning.

The deep learning model generated by the deep learning model generation unit 300 uses the image data obtained by converting the acceleration data as training data to determine whether the machine tool is abnormally operating, and determines whether the machine tool is abnormally operating based on the acceleration data. It is a model that can

Accordingly, the deep learning model is a model for determining whether the corresponding image belongs to an abnormal motion type of the machine tool by analyzing the image data corresponding to the acceleration data of the machine tool as input.

The abnormal motion determining unit 400 calls the deep learning model, inputs the target STFT image provided from the STFT processing unit 200 to the deep learning model, and determines the highest number of abnormal motion types classified through the deep learning model. The classified abnormal motion type may be determined as an abnormal motion type of corresponding acceleration data.

In the above, the configuration of the system for diagnosing an abnormal operation of a deep learning-based machine tool according to an embodiment of the present invention and the operation/function for each configuration have been described. Hereinafter, a diagnosis method using a system for diagnosing an abnormal operation of a machine tool based on deep learning according to an embodiment of the present invention will be described.

3 is a flowchart for explaining a method for diagnosing an abnormal operation of a deep learning-based machine tool according to an embodiment of the present invention, and the step-by-step operation shown in FIG. 3 is described with reference to FIGS. This can be done by the diagnostic system 1 .

The method for diagnosing abnormal motion of a machine tool based on deep learning according to FIG. 3 measures acceleration using an accelerometer 100 while moving a table T constituting an arbitrary machine tool C by driving a motor M, , Target used to create a deep learning model using the learning STFT image generated by STFT processing the learning acceleration data used to create the deep learning model, and to determine whether the machine tool (C) is operating abnormally It is possible to diagnose the abnormal operation of the machine tool (C) by determining the type of abnormal operation by inputting the target STFT image generated by STFT processing the acceleration data to the deep learning model.

1 to 3, while moving the table T of the machine tool C through the driving of the motor M, the deep learning model used to determine the abnormal operation of the machine tool C The learning acceleration used to generate may be measured using the accelerometer 100 (S300).

Then, the STFT processing unit 200 may generate a learning STFT image through STFT processing on the learning acceleration data (S310).

Thereafter, the deep learning model generation unit 300 may generate a deep learning model for determining an abnormal operation of the machine tool C using the training STFT image (S320).

In step S320, the deep learning model is an autoencoder-based model, and since the autoencoder-based model is a model that can be learned even if normal/abnormal data is imbalanced, unlike general classification machine learning, Suitable for judging abnormal operation.

After the deep learning model for determining the abnormal operation of the machine tool (C) is created, the table (T) of the machine tool (C) is moved through the driving of the motor (M), and the abnormal operation of the machine tool (C) is determined Target acceleration for may be measured using the accelerometer 100 (S330).

Then, the STFT processing unit 200 may generate a target STFT image through STFT processing on the target acceleration data (S340).

Thereafter, the abnormal motion determination unit 400 calls the deep learning model, inputs the target STFT image to the deep learning model, and selects the abnormal motion type classified by the highest number of times among the abnormal motion types classified through the deep learning model. It may be determined as an abnormal operation type of the corresponding acceleration data (S350).

Although the embodiments of the present invention have been described in more detail with reference to the accompanying drawings, the present invention is not necessarily limited to these embodiments, and may be variously modified and implemented without departing from the technical spirit of the present invention. . Therefore, the embodiments disclosed herein are not intended to limit the technical spirit of the present invention, but to explain, and the scope of the technical spirit of the present invention is not limited by these embodiments. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. The protection scope of the present invention should be construed according to the scope of the claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

1: Machine tool abnormal operation diagnosis system
100: accelerometer
200: STF processing unit
300: deep learning model generation unit
400: abnormal operation determination unit
500: database

Claims (4)

an accelerometer installed on a table of a machine tool and measuring an acceleration when the table moves;
an STFT processing unit generating an STFT image by performing Short Time Fourier Transform (STFT) processing on the acceleration data measured by the accelerometer;
a deep learning model generation unit for generating a deep learning model for determining an abnormal operation of the machine tool by using the STFT image from the STFT processing unit; and
A deep learning-based machine tool abnormal operation diagnosis system including an abnormal operation determination unit that calls the deep learning model and inputs the STFT image from the STFT processing unit to the deep learning model to determine the abnormal operation type of the corresponding acceleration data. According to claim 1,
The STFT processing unit provides a learning STFT image generated by STFT processing the learning acceleration data used to generate the deep learning model to the deep learning model generation unit, and determines abnormal operation of the machine tool. A system for diagnosing abnormal motion of a machine tool based on deep learning, wherein a target STFT image generated by STFT processing target acceleration data is provided to the abnormal motion determining unit. According to claim 1,
The deep learning model is an autoencoder-based model, a deep learning-based machine tool abnormal motion diagnosis system. According to claim 1,
Wherein the abnormal operation determining unit determines the abnormal operation type classified with the highest number of abnormal operation types classified through the deep learning model as the abnormal operation type of the corresponding acceleration data.

Images