KR102471830B1 - Apparatus and method for simultaneous monitoring of bearing and tool condition - Google Patents

Abstract

The present invention relates to an apparatus and method for simultaneous monitoring of bearing and tool conditions, a data measurement unit for measuring vibration data generated when a machine tool is operating, and extracting characteristics of vibration data generated under normal bearing and damaged bearing conditions. Bearing diagnosis model part that generates a bearing diagnosis model by configuring learning data, and tool condition diagnosis that extracts characteristics of vibration data generated in the machining process using normal and worn tools and configures learning data to create a tool condition diagnosis model A model unit and a monitoring unit configured to simultaneously monitor bearings and tools of the machine tool through the bearing diagnosis model and the tool condition diagnosis model based on the measured vibration data to diagnose the state of the bearings and the tool do.

Description

Device and method for simultaneous monitoring of bearing and tool condition {APPARATUS AND METHOD FOR SIMULTANEOUS MONITORING OF BEARING AND TOOL CONDITION}

The present invention relates to a bearing and tool condition monitoring technology, and more particularly, to a bearing and tool capable of simultaneously performing bearing and tool monitoring by collecting vibration data in real time in a machine tool processing process and diagnosing the condition of the bearing and tool. It relates to an apparatus and method for simultaneous state monitoring.

In general, damage to components in complex rotating machines may occur in unexpected situations due to noise or various factors. Failures of machine tool rotating bodies appear in tools and spindle bearings that process workpieces. Since bearings constitute an important part of rotating machinery, various data such as vibration and temperature were measured to predict or diagnose bearing conditions. The tool of a machine tool is a factor that determines the quality of a workpiece, and the state of the tool is diagnosed with data collected using various sensors (acceleration sensor, acoustic sensor, etc.) or the level of tool wear is measured using measuring equipment.

Existing machine tool monitoring systems use the spindle load indicated by NC to estimate the condition of the bearing, or the worker's judgment and the worn part of the tool are measured through a sensor, and the bearing and tool monitoring are individually performed to monitor the bearing and tool condition. Diagnosis takes a lot of time, which lowers the productivity of the machining process.

Korean Patent Registration No. 10-1957711 (2019.03.07) intelligently monitors and diagnoses the cutting state and diagnoses cutting conditions by utilizing a cutting characteristic map that maps the machining position and physical cutting characteristic values according to the machining time in the machining coordinate system. It relates to an intelligent CNC machine tool control system capable of controlling the CNC machine tool by using a cutting characteristic map that maps physical result values such as machining position and corresponding machining vibration and machining sound according to machining time on the CNC machining coordinate system. When cutting is performed, the cutting state monitoring function, the diagnosis function for the defective cutting state, etc., and the technology capable of changing and controlling defective and low-productive cutting conditions to high-quality and high-productive cutting conditions are performed. start about

Korean Patent Publication No. 10-2018-0024093 (2018.03.08) relates to a tool damage monitoring method of a machine tool based on a cutting load reflecting actual moving speed, information on tools, workpieces, machining routes and feed rates Calculating the arithmetic cutting load in consideration of, inputting the machining path and feed rate to the machine tool to derive the measured feed rate actually measured when machining the workpiece, correcting the arithmetic cutting load through the measured feed rate as a standard Calculating the cutting load, measuring the actual cutting load input to the tool when processing the workpiece through the tool, and determining whether the tool is damaged in preparation for the actual cutting load and the reference cutting load, Disclosed is a technique capable of increasing the accuracy of state monitoring.

Korean Patent Registration No. 10-1957711 (2019.03.07) Korean Patent Publication No. 10-2018-0024093 (2018.03.08)

An embodiment of the present invention provides an apparatus and method for simultaneous monitoring of bearing and tool conditions, which can simultaneously perform bearing and tool monitoring by collecting vibration data in real time in a machine tool machining process and diagnose the condition of bearings and tools. want to do

An embodiment of the present invention utilizes a spindle acceleration signal of a machine tool to generate a machine learning-based bearing and tool condition diagnosis model, respectively, and to use the bearing diagnosis model and tool condition diagnosis model in real time during the cutting process of the machine tool. And to provide a device and method for simultaneous monitoring of bearing and tool conditions that can improve productivity by diagnosing tool conditions.

An embodiment of the present invention attaches a 3-axis acceleration sensor to a spindle and applies the acceleration data measured during the machining process to a bearing diagnosis model and a tool condition diagnosis model so that bearing and tool monitoring can be performed simultaneously. It is intended to provide a device and method for monitoring.

Among the embodiments, an apparatus for simultaneously monitoring bearing and tool conditions includes a data measuring unit that measures vibration data generated when a machine tool is operating, extracts characteristics of vibration data generated under normal bearing and damaged bearing conditions, and learns data. A bearing diagnosis model unit that creates a bearing diagnosis model by configuring a tool condition diagnosis model unit that extracts characteristics of vibration data generated in the machining process using normal and worn tools and configures learning data to create a tool condition diagnosis model, and a monitoring unit for diagnosing states of the bearings and tools by simultaneously monitoring the bearings and tools of the machine tool through the bearing diagnosis model and the tool state diagnosis model based on the measured vibration data.

The data measurement unit may attach an acceleration sensor to the spindle of the machine tool and measure vibration according to the operation of the spindle through the acceleration sensor.

The bearing diagnosis model unit may separate frequency components from vibration data generated in normal bearings and damaged bearings through an empirical mode decomposition method to extract characteristic data for natural frequencies of bearing defects compared to normal data.

The bearing diagnosis model unit may select feature data extracted through a correlation function and generate a bearing diagnosis model for distinguishing between normal and abnormal bearings through machine learning using the selected feature data as an input.

The bearing diagnosis model unit clusters the extracted feature data through a correlation function to form a first cluster composed of feature data of normal bearings and a second cluster composed of feature data of each defect of damaged bearings, and the first cluster and the second cluster are formed. The feature data may be selected based on dissimilarity between clusters.

The bearing diagnosis model unit may calculate the dissimilarity according to the following [Equation Equation].

[mathematical expression]

Here, r _c1 is the radius of the cluster of normal data, r _c2 is the radius of the cluster of abnormal data, and D _c1,c2 is the distance between the centers of the two clusters.

The bearing diagnosis model unit may generate a bearing diagnosis model by inputting the selected feature data to a support vector machine.

The tool condition diagnosis model unit collects vibration data of a machining process using a worn tool and a normal tool through an acceleration sensor, extracts features for the collected vibration data, learns the extracted feature data as an input, and classifies normal and wear by learning the neural network. A tool condition diagnosis model can be created.

The tool condition diagnosis model unit may display the classified normal and worn conditions in a confusion matrix to confirm normal and worn classification rates of the tool.

Among the embodiments, a method for simultaneous monitoring of bearing and tool conditions includes the steps of extracting features based on vibrations generated in normal bearing and damaged bearing conditions and constructing learning data to generate a bearing diagnosis model; normal and worn tools; Extracting features based on vibration generated in the used machining process and constructing learning data to create a tool condition diagnosis model, measuring vibration data generated when the machine tool is running, and measuring the vibration data and diagnosing states of the bearings and tools by simultaneously monitoring the bearings and tools of the machine tool based on the bearing diagnosis model and the tool state diagnosis model.

The bearing diagnosis model generating step includes collecting vibration data generated in normal bearing and damaged bearing conditions, separating frequency components from the collected vibration data to extract characteristic data for the natural frequency of bearing defects compared to normal data, The method may include selecting extracted feature data for identifying abnormalities in the bearing through a correlation function, and generating a bearing diagnosis model by performing machine learning with the selected feature data.

In the bearing diagnostic model generation step, the dissimilarity between the first cluster consisting of characteristic data of normal bearings and the second cluster consisting of characteristic data of each defect of damaged bearings is calculated through the following equation in the characteristic data selection process, and the calculated dissimilarity is also calculated. may select feature data that satisfies a specific criterion.

[mathematical expression]

Here, r _c1 is the radius of the cluster of normal data, r _c2 is the radius of the cluster of abnormal data, and D _c1,c2 is the distance between the centers of the two clusters.

The tool state diagnosis model creation step includes collecting vibration data of a machining process using a worn tool and a normal tool, extracting features from the collected vibration data, and learning the extracted feature data as an input to a neural network to learn normal and wear conditions. It may include the step of generating a tool state diagnosis model by classifying the.

The disclosed technology may have the following effects. However, it does not mean that a specific embodiment must include all of the following effects or only the following effects, so it should not be understood that the scope of rights of the disclosed technology is limited thereby.

An apparatus and method for simultaneous monitoring of bearing and tool conditions according to an embodiment of the present invention collects vibration data in real time in a machine tool processing process, simultaneously monitors bearings and tools, and diagnoses the conditions of bearings and tools. .

An apparatus and method for simultaneous monitoring of bearing and tool conditions according to an embodiment of the present invention utilizes a spindle acceleration signal of a machine tool to generate a machine learning-based bearing and tool condition diagnosis model, respectively, and diagnose the bearing diagnosis model and tool condition Productivity can be improved by diagnosing the condition of bearings and tools in real time during the cutting process of machine tools through the model.

An apparatus and method for simultaneous monitoring of bearing and tool conditions according to an embodiment of the present invention attaches a 3-axis acceleration sensor to a spindle and applies acceleration data measured during a machining process to a bearing diagnosis model and a tool condition diagnosis model to monitor bearings. and tool monitoring can be performed simultaneously.

1 is a diagram illustrating a system for simultaneously monitoring bearing and tool conditions according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating a device for simultaneous monitoring of bearing and tool conditions in FIG. 1 .
FIG. 3 is a flowchart illustrating a process of generating a bearing diagnosis model of the bearing diagnosis model unit in FIG. 2 .
FIG. 4 is a flowchart illustrating a process of generating a tool state diagnosis model of the tool state diagnosis model unit in FIG. 2 .
5 is an exemplary view showing the configuration of a bearing diagnostic test apparatus according to an embodiment.
6 is an exemplary diagram illustrating types of defects of a bearing according to an exemplary embodiment.
7A and 7B are exemplary diagrams illustrating characteristics of vibration data according to normal and defective bearings according to an embodiment.
8 is an exemplary diagram illustrating a state in which characteristic data according to normal and defective bearings are displayed on a coordinate system according to an embodiment.
9 is an exemplary diagram illustrating clustering using a correlation function of feature data according to normal and defective bearings according to an embodiment.
10 is an exemplary view illustrating distribution of magnitudes of vibration data of a normal tool and a worn tool over time according to an exemplary embodiment.
11 is an exemplary diagram illustrating a neural network structure for a tool state diagnosis model according to an embodiment.
12 is an exemplary diagram illustrating a confusion matrix for classifying normal and worn conditions of a tool condition diagnosis model according to an exemplary embodiment.
13 is an exemplary diagram illustrating a screen for simultaneously monitoring bearing and tool conditions according to an exemplary embodiment.

Since the description of the present invention is only an embodiment for structural or functional description, the scope of the present invention should not be construed as being limited by the embodiments described in the text. That is, since the embodiment can be changed in various ways and can have various forms, it should be understood that the scope of the present invention includes equivalents capable of realizing the technical idea. In addition, since the object or effect presented in the present invention does not mean that a specific embodiment should include all of them or only such effects, the scope of the present invention should not be construed as being limited thereto.

Meanwhile, the meaning of terms described in this application should be understood as follows.

Terms such as "first" and "second" are used to distinguish one component from another, and the scope of rights should not be limited by these terms. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element.

It should be understood that when an element is referred to as being “connected” to another element, it may be directly connected to the other element, but other elements may exist in the middle. On the other hand, when an element is referred to as being "directly connected" to another element, it should be understood that no intervening elements exist. Meanwhile, other expressions describing the relationship between components, such as “between” and “immediately between” or “adjacent to” and “directly adjacent to” should be interpreted similarly.

Expressions in the singular number should be understood to include plural expressions unless the context clearly dictates otherwise, and terms such as “comprise” or “having” refer to an embodied feature, number, step, operation, component, part, or these. It should be understood that it is intended to indicate that a combination exists, and does not preclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

In each step, the identification code (eg, a, b, c, etc.) is used for convenience of explanation, and the identification code does not describe the order of each step, and each step clearly follows a specific order in context. Unless otherwise specified, it may occur in a different order than specified. That is, each step may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order.

The present invention can be implemented as computer readable code on a computer readable recording medium, and the computer readable recording medium includes all types of recording devices storing data that can be read by a computer system. . Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage devices. In addition, the computer-readable recording medium may be distributed to computer systems connected through a network, so that computer-readable codes may be stored and executed in a distributed manner.

All terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs, unless defined otherwise. Terms defined in commonly used dictionaries should be interpreted as consistent with meanings in the context of the related art, and cannot be interpreted as having ideal or excessively formal meanings unless explicitly defined in the present application.

1 is a diagram illustrating a system for simultaneously monitoring bearing and tool conditions according to an embodiment of the present invention.

Referring to FIG. 1 , a system 100 for simultaneously monitoring bearing and tool conditions may include a device 130 for monitoring the conditions of a bearing 111 and a tool 113 of a machine tool 110 .

The machine tool 110 is a machine for making various machines, and refers to a machine for processing such as metal cutting.

The device 130 for simultaneous monitoring of bearing and tool conditions is implemented as a server corresponding to a computer or program capable of diagnosing the conditions of the bearings 111 and tools 113 in real time by monitoring the cutting process of the machine tool 110. It can be. The device 130 for simultaneous monitoring of bearing and tool conditions is a bearing diagnosis model for diagnosing normality and damage of the bearing 111 based on vibration generated when the machine tool 110 is operating, and wear of the tool 113. A tool state diagnosis model for diagnosis may be created and the state of the bearing 111 and the tool 113 may be monitored based on the bearing diagnosis model and the tool state diagnosis model.

FIG. 2 is a block diagram illustrating a device for simultaneous monitoring of bearing and tool conditions in FIG. 1 .

Referring to FIG. 2 , the device 130 for simultaneous monitoring of bearing and tool conditions includes a data measurement unit 210, a bearing diagnosis model unit 230, a tool status diagnosis model unit 250, a monitoring unit 270, and a control unit. (290).

The data measurement unit 210 may measure vibration data generated when the machine tool 110 is operated. In one embodiment, the data measurement unit 210 may attach a 3-axis acceleration sensor to the spindle of the machine tool 110 to measure vibration according to spindle operation. Here, the data measurer 210 may collect vibration data for each direction of the X-axis, Y-axis, and Z-axis using a 3-axis acceleration sensor.

The bearing diagnosis model unit 230 may generate a bearing diagnosis model by extracting characteristics of vibration data generated under normal bearing and damaged bearing conditions and configuring learning data. In one embodiment, the bearing diagnosis model unit 230 may extract characteristics of vibration data measured using an empirical mode decomposition method.

The empirical mode decomposition method is a technique that applies the Hilbert Transform and can separate signals of time and amplitude based on frequency, which is called an intrinsic mode function. Abnormal data and normal data can be distinguished using intrinsic mode functions.

The bearing diagnostic modeling unit 230 may extract an intrinsic mode function by separating a frequency component from vibration data measured by applying an empirical mode decomposition method, and may extract feature data for the extracted intrinsic mode function. Here, the feature data is a natural frequency component of a bearing defect, and represents kurtosis, skewness, standard deviation, and average value compared to normal data. The bearing diagnosis model unit 230 may generate a bearing diagnosis model through machine learning by selecting feature data extracted through a selection criterion. In one embodiment, the bearing diagnosis model unit 230 may select feature data using a correlation function, and inputs the selected feature data to a support vector machine to perform bearing diagnosis for distinguishing between normal and abnormal bearings. model can be created.

The tool condition diagnosis model unit 250 may generate a tool condition diagnosis model by extracting characteristics of vibration data generated in a machining process using normal and worn tools and configuring learning data. In one embodiment, the tool condition diagnosis model unit 250 extracts feature data through a continuous wavelet transform technique for vibration data measured from a normal tool and vibration data measured from a worn tool, and features data of abnormal data compared to normal data. can be selected to create a deep learning-based tool condition diagnosis model that distinguishes normal tools from worn tools through deep learning using convolutional neural networks.

The monitoring unit 270 simultaneously monitors the bearing 111 and the tool 113 of the machine tool 110 through a bearing diagnosis model and a tool condition diagnosis model based on vibration data measured when the machine tool 110 is being processed. The state of the bearing 111 and the tool 113 can be quickly diagnosed by performing. In one embodiment, the monitoring unit 270 replaces the bearing 111 and the tool 113 at an appropriate time based on the diagnosis results of the bearing 111 and the tool 113 so that the productivity of the machine tool 110 is achieved. improvement can be provided.

The control unit 290 controls the overall operation of the device 130 for simultaneous monitoring of bearing and tool conditions, and includes the data measurement unit 210, the bearing diagnosis model unit 230, the tool status diagnosis model unit 250, and the monitoring unit. Control flow and data flow between (270) can be managed.

FIG. 3 is a flowchart illustrating a bearing diagnosis model generation process of the bearing diagnosis model unit in FIG. 2 .

In FIG. 3 , the bearing diagnosis model unit 230 collects vibration data generated under conditions of a normal bearing and a damaged bearing (step S310). The bearing diagnosis model unit 230 may collect vibration data by measuring vibration generated when the machine tool 110 is operated using an acceleration sensor. The bearing diagnosis model unit 230 may collect vibration data generated when the bearing 111 of the machine tool 110 is normal and vibration data generated when the bearing 111 has a defect. For example, as shown in FIG. 4 , the bearing diagnosis model unit 230 collects vibration data generated from the normal bearing 420 and the defective bearing 430 using the acceleration sensor 410 .

4 is an exemplary diagram illustrating a configuration of a bearing diagnostic test apparatus according to an exemplary embodiment, and FIG. 5 is an exemplary diagram illustrating defects of a bearing according to an exemplary embodiment.

As illustrated in FIG. 5, bearing defects occur in various ways, such as cage defects, smearing (stained state), and flaking (a phenomenon in which the surface is separated into thin pieces). Bearings change their vibration characteristics depending on the fault condition.

Returning to FIG. 3 again, the bearing diagnostic model unit 230 extracts features from the collected vibration data (step S330). The bearing diagnosis model unit 230 separates frequency components from vibration data and extracts feature data for bearing defects. In one embodiment, the bearing diagnostic modeling unit 230 extracts intrinsic mode functions of vibration data using an empirical mode decomposition method, which is a signal processing technique, and extracts feature data for the extracted intrinsic mode functions. Here, the characteristic data may include kurtosis, skewness, standard deviation, and average value. For example, as shown in FIG. 6A , the bearing diagnosis model unit 230 may extract an intrinsic mode function by separating vibration data of amplitude and time based on frequency using an empirical mode decomposition method.

6A and 6B are exemplary diagrams illustrating characteristics of intrinsic mode functions obtained by extracting vibration data according to normal and defective bearings using an empirical mode decomposition method according to an embodiment.

As illustrated in FIG. 6B , the intrinsic mode function of the vibration data of the ideal bearing has a larger amplitude than the intrinsic mode function of the normal vibration data. The intrinsic mode function of the vibration data of the faulty bearing shows different amplitudes depending on the type of defect.

7 is an exemplary diagram illustrating a state in which feature data is displayed on a coordinate system according to an exemplary embodiment.

As illustrated in FIG. 7 , the bearing diagnosis model unit 230 may display and classify characteristic data according to normal and defective types of bearings in a Cartesian coordinate system. For example, the bearing diagnosis model unit 230 may display and classify feature data extracted according to classification criteria of asymmetry, standard deviation, average, and root mean square on an orthogonal coordinate system consisting of an X axis and a Y axis. In one embodiment, the bearing diagnostic model unit 230 may combine a plurality of classification criteria into different types and display and classify feature data in an orthogonal coordinate system. For example, the bearing diagnostic model unit 230 may display and classify feature data in a Cartesian coordinate system based on the kurtosis value on the X axis and the asymmetry value on the Y axis. For another example, the bearing diagnostic model unit 230 may display and classify normal data and abnormal data of the bearing on the same Cartesian coordinate system based on the same classification criterion. In the case of FIG. 7 , bearing normal data (□) and defect-specific data (Flaking (○), Ball defect (▽), and Smearing (◇)) can be located in the same Cartesian coordinate system according to a plurality of classification criteria.

Returning to FIG. 3 again, the bearing diagnosis model unit 230 selects the extracted feature data to distinguish between normal and abnormal bearings (step S350). In one embodiment, the bearing diagnosis model unit 230 selects feature data through cluster analysis using a correlation function. As shown in FIG. 8 , the bearing diagnosis model unit 230 clusters the individual feature data classified by normal and defective bearings using a correlation function and selects the data based on the distance between the clusters.

8 is an exemplary diagram illustrating clustering of feature data using a data correlation function according to an embodiment.

As illustrated in FIG. 8 , the bearing diagnosis model unit 230 may form a first cluster consisting of characteristic data of normal bearings and a second cluster consisting of characteristic data of damaged bearings by using a correlation function.

The bearing diagnosis model unit 230 may calculate dissimilarity based on the distance between the first cluster and the second cluster through the following equation and select feature data that satisfies a specific criterion for the calculated dissimilarity.

[mathematical expression]

Here, r _c1 is the radius of the cluster of normal data, r _c2 is the radius of the cluster of abnormal data, and D _c1,c2 is the distance between the centers of the two clusters.

The bearing diagnostic model unit 230 determines that when the dissimilarity (DSN _F ) calculated through the above equation satisfies the condition greater than “1”, there is no overlapping feature between normal data and abnormal data of the bearing, which satisfies the condition. Select feature data.

The bearing diagnosis model unit 230 performs machine learning with the selected feature data to generate a bearing diagnosis model (step S370). In one embodiment, the bearing diagnosis model unit 230 may generate a bearing diagnosis model by inputting selected feature data to a support vector machine.

FIG. 4 is a flowchart illustrating a process of generating a tool state diagnosis model of the tool state diagnosis model unit in FIG. 2 .

Referring to FIG. 4 , the tool condition diagnosis model unit 250 collects tool vibration data (step S410). The tool state diagnosis model unit 250 may collect vibration data of a machining process using a worn tool and a normal tool through an acceleration sensor. As illustrated in FIG. 10 , the collected vibration data shows a size distribution over time in each direction of the X-axis, Y-axis, and Z-axis.

The tool state diagnosis model unit 250 extracts features from the collected vibration data (step S430). The tool state diagnosis model unit 250 may extract features by applying a continuous wavelet transform technique to vibration data. The tool state diagnosis model unit 250 generates a tool state diagnosis model by inputting the extracted feature data to a neural network (step S440). The tool condition diagnosis model unit 250 classifies normal and wear by learning by inputting feature data to the input layer of the neural network structure illustrated in FIG. 11 . The tool state diagnosis model unit 250 may display the classified normal and wear as a confusion matrix.

12 is an exemplary diagram illustrating a confusion matrix for classifying normal and worn conditions of a tool condition diagnosis model according to an exemplary embodiment.

Referring to FIG. 12 , normal and wear classification rates of tools can be confirmed in the confusion matrix. For example, as for the unclassified data, 47 out of 556 failed to be classified in the case of wear data and 66 out of 537 in the case of normal data.

13 is an exemplary diagram illustrating a screen for simultaneously monitoring bearing and tool conditions according to an exemplary embodiment.

Referring to FIG. 13 , the device 130 for simultaneous monitoring of bearing and tool conditions can simultaneously monitor vibration data measured by an acceleration sensor during a machining process through a bearing diagnosis model and a tool condition diagnosis model, and can simultaneously monitor bearing and tool conditions. can be diagnosed.

Although the above has been described with reference to preferred embodiments of the present invention, those skilled in the art will variously modify and change the present invention within the scope not departing from the spirit and scope of the present invention described in the claims below. You will understand that it can be done.

100: Simultaneous monitoring system for bearing and tool conditions
110: machine tool
111: bearing 113: tool
130: Device for simultaneous monitoring of bearing and tool condition
210: data measurement unit 230: bearing diagnosis model unit
250: tool condition diagnosis model unit 270: monitoring unit
290: control unit
510: acceleration sensor 520: normal bearing
530: defective bearing

Claims (13)

a data measurement unit for attaching a 3-axis acceleration sensor to a spindle and measuring vibration data of the spindle through the 3-axis acceleration sensor when the machine tool is in operation;
A bearing diagnosis model unit that extracts the average value, kurtosis, skewness, and standard deviation of vibration data generated under normal and damaged bearing conditions as feature data to form learning data and creates a bearing diagnosis model that distinguishes between normal and abnormal bearings. ;
A tool condition diagnosis model unit extracting features through continuous wavelet transform technique for vibration data generated in the machining process using normal and worn tools and constructing learning data to generate a tool condition diagnosis model; and
A monitoring unit for diagnosing the state of the bearing and the tool by simultaneously monitoring the bearing and tool of the machine tool through the bearing diagnosis model and the tool condition diagnosis model based on the measured vibration data,
The bearing diagnosis model unit
The extracted feature data is clustered using a correlation function to form a first cluster composed of feature data of normal bearings and a second cluster composed of feature data for each defect of damaged bearings;
If the dissimilarity between the first cluster and the second cluster is calculated through the following equation and the calculated dissimilarity satisfies the condition greater than “1”, it is the case where there is no overlap between normal data and abnormal data of the bearing, that condition Select feature data that satisfies
Performing machine learning with the selected feature data as input to generate the bearing diagnosis model,
[mathematical expression]

Here, r _c1 is the radius of the cluster of normal data, r _c2 is the radius of the cluster of abnormal data, D _c1,c2 is the distance between the centers of the two clusters,
The tool state diagnosis model unit
For simultaneous monitoring of bearing and tool conditions, characterized in that the tool condition diagnosis model is generated by selecting characteristic data of vibration data of a worn tool compared to vibration data of a normal tool and performing convolutional neural network learning to classify normal and wear. Device.
delete The method of claim 1, wherein the bearing diagnosis model unit
A device for simultaneous monitoring of bearing and tool conditions, characterized by extracting characteristic data on the natural frequency of bearing defects compared to normal data by separating frequency components from vibration data generated from normal bearings and damaged bearings through empirical mode decomposition.
delete delete delete The method of claim 1, wherein the bearing diagnosis model unit
An apparatus for simultaneous monitoring of bearing and tool conditions, characterized in that for generating a bearing diagnostic model by inputting the selected feature data into a support vector machine.
delete The method of claim 1, wherein the tool state diagnosis model unit
A device for simultaneous monitoring of bearing and tool conditions, characterized in that the normal and worn classification rates of the tool are confirmed by displaying the classified normal and wear as a confusion matrix.
Extracting features based on vibrations generated in normal bearing and damaged bearing conditions and constructing learning data to generate a bearing diagnosis model;
Generating a tool condition diagnosis model by extracting features through a continuous wavelet transform technique based on vibrations generated in a machining process using normal and worn tools and constructing learning data;
Measuring vibration data generated when the machine tool is operated; and
Diagnosing the state of the bearing and the tool by simultaneously monitoring the bearing and tool of the machine tool through the bearing diagnosis model and the tool condition diagnosis model based on the measured vibration data,
The bearing diagnosis model generation step is
Collecting vibration data generated in normal bearing and damaged bearing conditions;
Extracting characteristic data for the natural frequency of the bearing defect compared to normal data by separating frequency components from the collected vibration data;
The extracted feature data for identifying the abnormality of the bearing is clustered through a correlation function to form a first cluster composed of feature data of normal bearings and a second cluster composed of feature data of each defect of damaged bearings. Calculating dissimilarity between the two clusters through the following equation and selecting feature data for which the calculated dissimilarity satisfies a specific criterion; and
Generating a bearing diagnosis model by performing machine learning with the selected feature data,
[mathematical expression]

Here, r _c1 is the radius of the cluster of normal data, r _c2 is the radius of the cluster of abnormal data, D _c1,c2 is the distance between the centers of the two clusters,
The tool condition diagnosis model creation step is
For simultaneous monitoring of bearing and tool conditions, characterized by generating the tool condition diagnosis model that classifies normal and wear by selecting characteristic data of vibration data of a worn tool compared to vibration data of a normal tool and performing convolutional neural network learning. Way.
delete delete delete

Images