Tool life prediction device, tool life prediction method, and machine tool

Abstract

To provide a technique capable of predicting a residual life of a tool even when a working condition is changed.SOLUTION: A tool life prediction device includes: a signal acquisition part for acquiring an output signal from a sensor for detecting a physical quantity generated by bringing each of a plurality of blades into contact with workpiece cut by rotation of a tool having the plurality of blades; a self-correlation coefficient calculation part for calculating a self-correlation coefficient of the output signal; a variation degree calculation part for calculating a variation degree between a plurality of partial waveforms which constitute the waveforms of the self-correlation coefficient and correspond to the plurality of blades; and a residual life prediction part for predicting the residual line of the tool, on the basis of an index of the variation.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a tool life prediction device, a tool life prediction method, and a machine tool.

Background Art

[0002] Patent Document 1 discloses a technique for preventing a tool from being damaged during cutting and causing a processing defect by changing processing conditions when a peak value of vibration acceleration measured near the tool during cutting exceeds a threshold value.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] [[ID=3 forty]] In the above-described technique, the tool is prevented from being damaged during cutting by changing the processing conditions during cutting. In order to more reliably prevent the tool from being damaged during cutting, it is preferable to predict the remaining life of the tool. For example, as the remaining life of the tool decreases, vibrations, cutting sounds, etc., generated when the tool edge contacts the workpiece increase, so it is conceivable to predict the remaining life of the tool based on the magnitude of the vibration and the magnitude of the cutting sound. However, if the processing conditions are different, the above-described magnitudes of vibration and cutting sound are also different. Therefore, for example, when a tool is used in an environment where the processing conditions are frequently changed, such as in a multi-variety small-lot production line, it is difficult to predict the remaining life of the tool based on the magnitude of the vibration and the magnitude of the cutting sound as described above.

Means for Solving the Problems

[0005] The present disclosure can be realized in the following forms.

[0006] (1) According to a first embodiment of the present disclosure, a tool life prediction device is provided. This tool life prediction device includes: a signal acquisition unit that acquires an output signal from a sensor that detects a physical quantity caused by each of the multiple blades contacting a workpiece that is being cut by the rotation of a tool having multiple blades; an autocorrelation coefficient calculation unit that calculates the autocorrelation coefficient of the output signal; a variation degree calculation unit that calculates the degree of variation among a plurality of partial waveforms that constitute the waveform of the autocorrelation coefficient and which correspond to the plurality of blades; and a remaining life prediction unit that predicts the remaining life of the tool based on the degree of variation. In this type of tool life prediction device, the remaining life prediction unit predicts the remaining life of the tool based on the degree of variation between multiple partial waveforms. As wear progresses on each cutting edge of the tool, the degree of variation in wear of each cutting edge increases, and the degree of variation between multiple partial waveforms increases. The degree of variation between multiple partial waveforms changes with increasing machining time, but does not change with changes in machining conditions. Therefore, the remaining life prediction unit can predict the remaining life of the tool even if the machining conditions are changed. (2) In the tool life prediction device of the above form, the variation degree calculation unit may calculate the variation degree using the peak values ​​of the plurality of partial waveforms. This type of tool life prediction device reduces the burden of calculating the degree of variation compared to a system that calculates the degree of variation across the entire partial waveform. (3) In the tool life prediction device of the above form, the variation degree calculation unit may derive an approximate straight line of the peak values ​​of the plurality of partial waveforms, and calculate the difference between the peak values ​​and the approximate straight line as the variation degree. With this type of tool life prediction device, the greater the variation between multiple partial waveforms, the greater the difference between the peak value and the approximate straight line, allowing for a more accurate representation of the degree of variation. Therefore, the accuracy of remaining tool life prediction can be improved. (4) In the tool life prediction device of the above form, the remaining life prediction unit may predict the remaining life based on the difference between a predetermined threshold and the degree of variation. This type of tool life prediction device can predict the remaining machining time until the degree of variation reaches a threshold, which can be used to determine the remaining tool life. (5) The tool life prediction device of the above form may include a notification unit that notifies when the remaining life predicted by the remaining life prediction unit falls below a predetermined value. This type of tool life prediction device can notify the user when the remaining tool life is running low, thus preventing machining defects caused by the tool reaching the end of its lifespan during cutting operations. (6) A second embodiment of the present disclosure provides a tool life prediction method. This tool life prediction method includes: a signal acquisition step of acquiring an output signal from a sensor that detects a physical quantity caused by each of the multiple cutting edges contacting a workpiece that is being cut by the rotation of a tool having multiple cutting edges; an autocorrelation coefficient calculation step of calculating the autocorrelation coefficient of the output signal; a variation degree calculation step of calculating the variation degree among a plurality of partial waveforms that constitute the waveform of the autocorrelation coefficient and correspond to the plurality of cutting edges; and a remaining life prediction step of predicting the remaining life of the tool based on the variation degree. According to this type of tool life prediction method, the remaining tool life is predicted in the remaining life prediction process based on the degree of variation between multiple partial waveforms. As wear progresses on each cutting edge of the tool, the degree of variation in wear of each cutting edge increases, and the degree of variation between multiple partial waveforms increases. The degree of variation between multiple partial waveforms changes with increasing machining time, but does not change with changes in machining conditions. Therefore, even if the machining conditions are changed, the remaining tool life can be predicted. (7) A third embodiment of the present disclosure provides a machine tool. The machine tool comprises a spindle device that rotates the tool mounted on the spindle, and has a spindle on which a tool having a plurality of cutting edges is mounted; a moving device that moves the workpiece fixed to the table relative to the tool mounted on the spindle, and has a table on which a workpiece to be cut by the rotation of the tool is fixed; a control device that controls the spindle device and the moving device; and a sensor that detects a physical quantity caused by each of the plurality of cutting edges contacting the workpiece. The control device comprises a signal acquisition unit that acquires an output signal from the sensor; an autocorrelation coefficient calculation unit that calculates an autocorrelation coefficient of the output signal; a variation degree calculation unit that calculates the degree of variation between a plurality of partial waveforms that constitute the waveform of the autocorrelation coefficient and correspond to the plurality of cutting edges; and a remaining life prediction unit that predicts the remaining life of the tool based on the degree of variation. In this type of machine tool, the remaining tool life prediction unit predicts the remaining tool life based on the degree of variation between multiple partial waveforms. As wear progresses on each cutting edge of the tool, the degree of variation in wear of each cutting edge increases, and the degree of variation between multiple partial waveforms increases. The degree of variation between multiple partial waveforms changes with increasing machining time, but does not change with changes in machining conditions. Therefore, the remaining tool life prediction unit can predict the remaining tool life even if the machining conditions are changed. This disclosure can also be implemented in various forms other than tool life prediction devices, tool life prediction methods, and machine tools. For example, it can be implemented in the form of a control device for a machine tool. [Brief explanation of the drawing]

[0007] [Figure 1] A perspective view showing the schematic configuration of a machine tool according to the first embodiment. [Figure 2] An explanatory diagram showing the functional configuration of the control device of the first embodiment. [Figure 3] A cross-sectional view showing the blade of a tool. [Figure 4] A flowchart illustrating the process of predicting tool life. [Figure 5] The first explanatory diagram shows the waveform of the sensor's output signal. [Figure 6] A second explanatory diagram showing the waveform of the sensor's output signal. [Figure 7] An explanatory diagram showing the autocorrelation function of the sensor's output signal. [Figure 8] A schematic diagram illustrating the predicted remaining lifespan of a tool. [Modes for carrying out the invention]

[0008] A. First Embodiment: Figure 1 is a perspective view showing the schematic configuration of the machine tool 11 in the first embodiment. In this embodiment, the machine tool 11 is a vertical machining center. The machine tool 11 has three mutually orthogonal coordinate axes, namely the X, Y, and Z axes. In this embodiment, the X axis is a coordinate axis along the left-right direction of the machine tool 11, the Z axis is a coordinate axis along the front-back direction of the machine tool 11, and the Y axis is a coordinate axis along the up-down direction of the machine tool 11.

[0009] The machine tool 11 comprises a spindle unit 100, a moving device 200, a sensor 300, and a control device 400. The spindle unit 100 comprises a spindle 110, a spindle motor 115, and a rotation angle detector 120. A tool TL is mounted on the spindle 110. As will be described later, the tool TL has multiple cutting edges at its tip. In this embodiment, the tool TL is an end mill. For the spindle motor 115, for example, a servo motor or a built-in motor can be used. For the rotation angle detector 120, for example, a rotary encoder can be used.

[0010] The spindle unit 100 rotates the spindle 110 around a rotation axis RX parallel to the Y-axis using the spindle motor 115. As the spindle 110 rotates, the tool TL mounted on the spindle 110 also rotates around the rotation axis RX. The rotation angle detector 120 detects the rotation angle of the spindle 110, or in other words, the rotation angle of the tool TL. The signal representing the rotation angle detected by the rotation angle detector 120 is transmitted to the control device 400.

[0011] The moving device 200 includes a saddle 220 and a table 230. A workpiece WK is fixed to the table 230. The moving device 200 relatively moves the workpiece WK fixed to the table 230 with respect to the tool TL mounted on the spindle 110. As will be described later, the workpiece WK is machined by the tool TL.

[0012] In the present embodiment, a first rail 215 is provided on the upper surface of the bed 210 along the Z-axis. The saddle 220 is disposed on the first rail 215. A second rail 225 is provided on the upper surface of the saddle 220 along the X-axis. The table 230 is disposed on the second rail 225. The bed 210 and the saddle 220 are respectively provided with a servo motor and a ball screw (not shown). While guiding the saddle 220 by the first rail 215, the bed 210 moves the saddle 220 along the Z-axis by the servo motor and the ball screw. Along with the movement of the saddle 220, the table 230 and the workpiece WK move. While guiding the table 230 by the second rail 225, the saddle 220 moves the table 230 along the X-axis by the servo motor and the ball screw. Along with the movement of the table 230, the workpiece WK moves.

[0013] In the present embodiment, the column 240 is fixed to the upper surface of the bed 210. A third rail 245 is provided on the side surface of the column 240 along the Y-axis. The spindle device 100 is connected to the third rail 245. The column 240 is provided with a servo motor and a ball screw (not shown). While guiding the spindle device 100 by the third rail 245, the column 240 moves the spindle device 100 along the Y-axis by the servo motor and the ball screw.

[0014] The sensor 300 detects a physical quantity generated when each cutting edge provided on the tool TL cuts into the workpiece WK. The physical quantity generated when each cutting edge provided on the tool TL cuts into the workpiece WK means, for example, the acceleration of the vibration of the tool TL or the workpiece WK, the cutting sound, or the load current of the spindle motor 115 that rotates the tool TL. The cutting sound includes not only audible sound but also ultrasonic waves. It is preferable that a microphone or an ultrasonic microphone for detecting the cutting sound is used for the sensor 300. Since the ultrasonic microphone is less affected by ambient noise, it is particularly preferable that the ultrasonic microphone is used as the sensor 300. An acceleration sensor for detecting the acceleration of the vibration of the tool TL or the workpiece WK or an AE sensor for detecting AE waves may be used for the sensor 300. When the tool TL is large and rotates at a low speed, a current sensor for detecting the load current of the spindle motor 115 that rotates the tool TL may be used for the sensor 300. In the present embodiment, the sensor 300 is an acceleration sensor that detects the acceleration of the vibration of the tool TL generated when each cutting edge provided on the tool TL cuts into the workpiece WK, and is provided on the spindle 110. A signal representing the acceleration detected by the sensor 300 is transmitted to the control device 400. In the present embodiment, a filter circuit 350 for reducing noise included in the signal output from the sensor 300 is provided between the sensor 300 and the control device 400. For the filter circuit 350, for example, a high-pass filter that passes frequency components higher than the tool rotation frequency described later and blocks other frequency components is used. Note that, instead of the high-pass filter, for example, a band-pass filter that passes frequency components in the range from the tool rotation frequency to the cutting edge passing frequency described later and blocks other frequency components may be used for the filter circuit 350.

[0015] The control device 400 is configured as a computer including a CPU 401, a memory 402, and an input / output interface 403. Further, in the present embodiment, the control device 400 includes a display device 405. The display device 405 is configured of, for example, a liquid crystal display.

[0016] Figure 2 is an explanatory diagram showing the functional configuration of the control device 400 in this embodiment. In this embodiment, the control device 400 includes an NC control unit 410, a signal acquisition unit 420, an autocorrelation coefficient calculation unit 430, a variation degree calculation unit 440, a remaining life prediction unit 450, a notification unit 460, and a learning unit 470. The NC control unit 410, the signal acquisition unit 420, the autocorrelation coefficient calculation unit 430, the variation degree calculation unit 440, the remaining life prediction unit 450, the notification unit 460, and the learning unit 470 are each implemented in software by the CPU 401 executing a computer program stored in memory 402.

[0017] The NC control unit 410 controls the spindle unit 100 and the moving device 200 to perform cutting on the workpiece WK with the tool TL. In this embodiment, the NC control unit 410 rotates the tool TL mounted on the spindle 110 by driving the spindle motor 115 provided on the spindle unit 100, and drives each servo motor provided on the moving device 200 to move the workpiece WK, which is fixed to the table 230, relative to the tool TL in a direction perpendicular to the rotation axis RX of the tool TL, thereby performing cutting on the side surface of the workpiece WK with the tool TL. In this embodiment, the NC control unit 410 counts the cutting time by the tool TL and stores it in the memory 402.

[0018] The signal acquisition unit 420 acquires the output signal from the sensor 300 during cutting. The autocorrelation coefficient calculation unit 430 calculates the autocorrelation coefficient of the output signal from the sensor 300 acquired by the signal acquisition unit 420. As described later, the waveform of the autocorrelation coefficient calculated by the autocorrelation coefficient calculation unit 430 is composed of multiple partial waveforms corresponding to the multiple cutting edges of the tool TL. The variation degree calculation unit 440 calculates the variation degree between the multiple partial waveforms. The remaining life prediction unit 450 predicts the remaining life of the tool TL using the variation degree calculated by the variation degree calculation unit 440. The notification unit 460 notifies the user of information regarding the remaining life of the tool TL according to the prediction result of the remaining life prediction unit 450. The learning unit 470 performs machine learning on the learning model MD used for predicting the remaining life and stores the learned learning model MD in the memory 402. The control device 400 is sometimes referred to as the tool life prediction device.

[0019] Figure 3 is a cross-sectional view showing the four blades B1 to B4 provided at the tip of the tool TL. As described above, in this embodiment, the tool TL is an end mill. In this embodiment, four blades B1 to B4 are provided at the tip of the tool TL. Each blade B1 to B4 is provided at equal intervals on the side surface of the tool TL along the circumferential direction CD centered on the central axis CL of the tool TL. Note that the number of blades B1 to B4 on the tool TL is not limited to four; two or more are acceptable.

[0020] In this embodiment, the tool TL rotates clockwise around its central axis CL as shown in Figure 3. In the following description, the time it takes for the tool TL to complete one rotation is referred to as the tool rotation period. The time interval from when one of the multiple cutting edges B1 to B4 of the tool TL cuts into the workpiece WK until the adjacent cutting edge behind the cutting edge in the direction of rotation of the tool TL cuts into the workpiece WK is called the cutting edge passage period. When the tool TL is rotating at a constant speed, the cutting edge passage period can be calculated by dividing the tool rotation period by the number of cutting edges of the tool TL. The tool rotation period can be obtained by the rotation angle detector 120.

[0021] Each cutting edge B1 to B4 of the tool TL wears down as it repeatedly cuts the workpiece WK. Wear here means that the cutting edge of the tool TL wears down and its cutting performance decreases. In this embodiment, the spindle unit 100 rotates the tool TL, and the moving device 200 moves the workpiece WK relative to the rotating tool TL along a direction perpendicular to the rotation axis RX of the tool TL, thereby performing cutting on the side surface of the workpiece WK with the tool TL. During cutting, each cutting edge B1 to B4 of the tool TL successively cuts into the side surface of the workpiece WK. At the timing when the cutting edge first cuts into the workpiece WK, a greater force is applied to the cutting edge compared to the timing when the cutting edge cuts into the workpiece WK later, so the cutting edge that cuts into the workpiece WK first wears down or gets damaged more than the other cutting edges. Since the cutting edge that cuts into the workpiece WK first is not the same each time, the wear and damage to each cutting edge B1 to B4 of the tool TL does not progress evenly, resulting in variations in the wear and damage to each cutting edge B1 to B4. As tool TL wears down or becomes damaged, the degree of wear and damage to each blade B1 to B4 increases. Therefore, in this embodiment, tool TL is considered to have reached the end of its lifespan when the degree of wear and damage to each blade B1 to B4 exceeds a predetermined size, and the desired cutting performance can no longer be ensured.

[0022] Figure 4 is a flowchart showing the contents of the tool life prediction process performed by the control device 400 in this embodiment. Figure 5 is a first explanatory diagram showing the waveform of the output signal SG output from the sensor 300 during cutting. Figure 6 is a second explanatory diagram showing the waveform of the output signal SG. Figure 7 is an explanatory diagram showing the autocorrelation function of the output signal SG. In Figures 5 to 6, the horizontal axis represents time, and the vertical axis represents the value of the output signal from the sensor 300. In Figure 7, the horizontal axis represents the lag, which will be described later, and the vertical axis represents the autocorrelation coefficient, which will be described later.

[0023] In this embodiment, the tool life prediction process shown in Figure 4 is performed by the CPU 401 of the control device 400 while cutting is being performed by the machine tool 11. When the tool life prediction process is started, first, in step S110, the signal acquisition unit 420 acquires the output signal SG output from the sensor 300 during cutting. Figure 5 shows the waveform of the output signal SG. In this embodiment, the output signal SG changes in accordance with the change in the acceleration of vibration occurring in the tool TL. The signal acquisition unit 420 acquires the output signal SG obtained by measuring the vibration acceleration for a period of time sufficiently longer than the tool rotation period.

[0024] In step S120 of Figure 4, the autocorrelation coefficient calculation unit 430 extracts a portion of the output signal SG acquired in step S110. The autocorrelation coefficient calculation unit 430 extracts a portion of the output signal SG in which the amplitude and frequency are stable. In Figure 5, since the amplitude and frequency of the output signal SG are stable in the interval from time t1 to time t2, the autocorrelation coefficient calculation unit 430 extracts the interval from time t1 to time t2 of the output signal SG. Figure 6 shows the waveform of the output signal SG in the interval from time t1 to time t2. If an interval with unstable amplitude and frequency of the output signal SG is extracted, such as the interval from time t3 to time t4, the accuracy of the degree of variation, which will be described later, will decrease, so it is preferable for the autocorrelation coefficient calculation unit 430 not to extract such intervals.

[0025] In step S130 of Figure 4, the autocorrelation coefficient calculation unit 430 derives the autocorrelation function of the output signal SG. The autocorrelation function is a function that represents the similarity between the waveform before and after a shift when the waveform of a signal is shifted along the time axis. The amount of shift is called the lag. The degree of similarity is called the autocorrelation coefficient. The autocorrelation coefficient is a value between -1.0 and 1.0. The larger the autocorrelation coefficient, the higher the similarity between the waveform before and after the shift. The autocorrelation function can be expressed using the following equation (1).

number

[0026] In step S140 of Figure 4, the variation degree calculation unit 440 detects the peak PK of the autocorrelation coefficient waveform W. As shown in Figure 7, when the magnitude of the lag τ corresponds to an integer multiple of the tool rotation period Tr, the peaks of the waveform of the output signal SG generated by each cutting edge B1 to B4 of the tool TL cutting into the workpiece WK overlap with the peaks of the waveform of the output signal SG generated by each cutting edge B1 to B4 itself cutting into the workpiece WK, so a peak PK appears in the autocorrelation coefficient waveform W.

[0027] In step S150 of Figure 4, the variation degree calculation unit 440 derives an approximate straight line for each peak PK detected in step S140. As shown in Figure 7, in this embodiment, the variation degree calculation unit 440 derives an approximate straight line for each peak PK using the least squares method.

[0028] In step S160 of Figure 4, the variation degree calculation unit 440 calculates the variation degree VB of each partial waveform PW that constitutes the waveform W of the autocorrelation coefficient, in other words, the variation degree VB of the wear and damage of each cutting edge B1 to B4 of the tool TL. In this embodiment, the variation degree calculation unit 440 calculates the difference between the peak value and the approximate straight line for each peak PK of each partial waveform PW. The variation degree calculation unit 440 obtains the maximum value among the differences between each peak value and the approximate straight line as the variation degree VB. In Figure 7, the variation degree calculation unit 440 obtains the difference between the peak value and the approximate straight line as the variation degree VB when the size of the lag τ corresponds to the tool rotation period Tr.

[0029] As the variation in wear and damage of each cutting edge B1 to B4 of the tool TL increases, the variation in the vibration acceleration of the tool TL, the vibration acceleration of the workpiece WK, the cutting noise, and the load on the spindle motor 115 increases when each cutting edge B1 to B4 cuts into the workpiece WK. In this embodiment, since the output signal SG of the sensor 300 corresponds to the vibration acceleration of the tool TL, as the variation in wear and damage of each cutting edge B1 to B4 increases, the variation in the output signal SG when each cutting edge B1 to B4 cuts into the workpiece WK increases. Therefore, the maximum value of the difference between each peak value and the approximate straight line can be used as an indicator to represent the magnitude of the variation in wear and damage of each cutting edge B1 to B4.

[0030] In step S170, the remaining life prediction unit 450 predicts the remaining life of the tool TL using the variability degree VB calculated in step S160. In this embodiment, the remaining life prediction unit 450 predicts the remaining machining time until the variability degree VB of the tool TL reaches a threshold as the remaining life of the tool TL. For example, the variability degree VB of the tool TL when the cutting performance of the workpiece WK falls outside the acceptable range can be determined by a prior test, and this variability degree VB can be used as the threshold.

[0031] In this embodiment, the remaining life prediction unit 450 predicts the remaining life of the tool TL using a trained learning model MD stored in the memory 402. For example, a neural network can be used for the learning model MD. The learning model MD is configured to output a prediction line representing the increasing trend of the future variation degree VB when the current variation degree VB is input, by learning pairs of past variation degree VB and cumulative machining time as training data. Based on the prediction line output from the learning model MD, the remaining life prediction unit 450 predicts the remaining machining time until the variation degree VB of the tool TL reaches a threshold. In other embodiments, the learning model MD may be configured to output the remaining life when the variation degree VB is input, by learning pairs of variation degree VB and remaining life obtained from prior tests as training data.

[0032] In step S180, the remaining life prediction unit 450 outputs the prediction result. In this embodiment, the remaining life prediction unit 450 displays the prediction result on the display device 405 and transmits it to the notification unit 460. Figure 8 shows the prediction result displayed on the display device 405. In this embodiment, the notification unit 460 notifies the user that the tool TL is nearing the end of its life when the remaining life of the tool TL predicted by the remaining life prediction unit 450 falls below a predetermined value. In this embodiment, the notification unit 460 notifies the user that the tool TL is nearing the end of its life by displaying a message on the display device 405. In other embodiments, a buzzer for notifying that the tool TL is nearing the end of its life may be provided on the machine tool 11, and the notification unit 460 may notify the user that the tool TL is nearing the end of its life by sounding the buzzer. Alternatively, the machine tool 11 may be equipped with a lamp to indicate that the tool TL is nearing the end of its lifespan, and 6 may notify the user that the tool TL is nearing the end of its lifespan by illuminating the lamp.

[0033] After that, the CPU 401 of the control device 400 terminates this process. The method executed by the tool life prediction process is sometimes called the tool life prediction method. Step S110 is sometimes called the signal acquisition process. Steps S120 to S130 are sometimes called the autocorrelation coefficient calculation process. Steps S140 to S160 are sometimes called the variability degree calculation process. Step S170 is sometimes called the remaining life prediction process. Step S180 is sometimes called the notification process.

[0034] In the machine tool 11 of this embodiment described above, the remaining life prediction unit 450 predicts the remaining life of the tool TL based on the degree of variation VB of each partial waveform PW that constitutes the waveform W of the autocorrelation coefficient, in other words, the degree of variation VB of the wear and damage of each cutting edge B1 to B4 of the tool TL. The degree of variation VB is a value calculated using the autocorrelation function of the output signal SG output from the sensor 300 that detects the vibration of the workpiece WK during cutting, and changes with increasing machining time, but does not change with changes in machining conditions. Therefore, the remaining life prediction unit 450 can predict the remaining life of the tool TL even if the machining conditions are changed.

[0035] Furthermore, in this embodiment, the variation degree calculation unit 440 calculates the variation degree VB using the peak values ​​of multiple partial waveforms PW. Therefore, compared to the form in which the variation degree is calculated for the entire waveform of the partial waveforms PW, the burden on the CPU 401 and memory 402 can be reduced.

[0036] Furthermore, in this embodiment, the variation degree calculation unit 440 derives an approximate straight line of the peak values ​​of multiple partial waveforms PW, and calculates the difference between the peak value and the approximate straight line as the variation degree VB. The greater the variation in wear and damage of each cutting edge B1 to B4 of the tool TL, the greater the difference between the peak value and the approximate straight line. Therefore, by using the variation degree VB calculated by the method described above, the variation in wear and damage of each cutting edge B1 to B4 can be detected with high accuracy.

[0037] Furthermore, in this embodiment, the remaining life prediction unit 450 predicts the remaining machining time until the degree of variation VB reaches a threshold as the remaining life of the tool TL. Therefore, the remaining life of the tool TL can be easily predicted. In particular, in this embodiment, the remaining life is predicted using a trained model MD that has been trained using past performance as training data, so the remaining life of the tool TL can be predicted with high accuracy.

[0038] Furthermore, in this embodiment, the control device 400 includes a notification unit 460 that notifies the user when the remaining life predicted by the remaining life prediction unit 450 falls below a predetermined value. Therefore, when the remaining life becomes short, the user is notified that the remaining life has become short, thus preventing the tool TL from reaching the end of its life during cutting and causing machining defects in the workpiece WK.

[0039] B. Other embodiments: (B1) The machine tool 11 in the first embodiment described above is a vertical machining center. In contrast, the machine tool 11 may be a horizontal machining center or an NC milling machine, for example, instead of a vertical machining center. The tool TL may be a milling tool other than an end mill, such as a flat milling cutter, instead of an end mill.

[0040] (B2) In the first embodiment described above, the control device 400 includes an NC control unit 410, a signal acquisition unit 420, an autocorrelation coefficient calculation unit 430, a variation degree calculation unit 440, a remaining life prediction unit 450, a notification unit 460, and a learning unit 470. In contrast, the signal acquisition unit 420, the autocorrelation coefficient calculation unit 430, the variation degree calculation unit 440, the remaining life prediction unit 450, the notification unit 460, and the learning unit 470 may be provided in a computer connected to the control device 400. In this case, the computer, rather than the control device 400, is referred to as the tool life prediction device.

[0041] (B3) In the first embodiment described above, the autocorrelation coefficient calculation unit 430 may derive the autocorrelation function by equation (2) below instead of equation (1) above. The function ACF(τ) expressed by equation (2) below is, strictly speaking, a function similar to the autocorrelation function, but here we will also refer to the function ACF(τ) as the autocorrelation function. Strictly speaking, it is a function that resembles autocorrelation. ACF(τ) = IFFT(|FFT(x(t))| 2 ) ···(2) Here, ACF(τ) is the autocorrelation function of the output signal SG, where τ is the lag. x(t) is the output signal SG, where t is time. FFT() is the Fast Fourier Transform (FFT) in parentheses. IFFT() is the Inverse Fast Fourier Transform (IFFT) in parentheses.

[0042] (B4) In the first embodiment described above, the remaining life prediction unit 450 predicts the remaining life of the tool TL using a learned learning model MD. In contrast, the remaining life prediction unit 450 may predict the remaining life of the tool TL without using the learning model MD. For example, an approximate straight line of a point cloud representing the relationship between the past variability VB and the cumulative machining time may be derived using the least squares method, and the remaining life may be calculated using the approximate straight line.

[0043] (B5) In the first embodiment described above, the variability calculation unit 440 calculates the variability VB using the peak value of the autocorrelation coefficient waveform. Alternatively, the variability calculation unit 440 may calculate the variability VB using the autocorrelation coefficient at a position shifted from the peak PK.

[0044] This disclosure is not limited to the embodiments described above, and can be implemented in various configurations without departing from its spirit. For example, the technical features in the embodiments corresponding to the technical features in each form described in the summary of the invention can be replaced or combined as appropriate in order to solve some or all of the above-described problems, or to achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be deleted as appropriate. [Explanation of Symbols]

[0045] 11...Machine tool, 100...Spindle unit, 110...Spindle, 115...Spindle motor, 120...Rotation angle sensor, 200...Movement device, 210...Bed, 215...First rail, 220...Saddle, 225...Second rail, 230...Table, 240...Column, 245...Third rail, 300...Sensor, 350...Filter circuit, 400...Control device, 401...CPU, 402...Memory, 403...Input / output interface, 405...Display device, 410...NC control unit, 420...Signal acquisition unit, 430...Autocorrelation coefficient calculation unit, 440...Variation degree calculation unit, 450...Remaining life prediction unit, 460...Notification unit, 470...Learning unit, B1~B4...Blade, TL...Tool, WK...Workpiece

Claims

1. A tool life prediction device, A signal acquisition unit acquires output signals from a sensor that detects a physical quantity generated when each of the multiple blades of a tool having multiple blades comes into contact with a workpiece being cut by the rotation of the tool, An autocorrelation coefficient calculation unit that calculates the autocorrelation coefficient of the output signal, A variation degree calculation unit that calculates the degree of variation among a plurality of partial waveforms that constitute the waveform of the autocorrelation coefficient, and which corresponds to a plurality of blades, A remaining life prediction unit predicts the remaining life of the tool based on the degree of variation, A tool life prediction device equipped with the following features.

2. A tool life prediction device according to claim 1, The variation degree calculation unit is a tool life prediction device that calculates the variation degree using the peak values ​​of the plurality of partial waveforms.

3. A tool life prediction device according to claim 2, The tool life prediction device comprises a variation degree calculation unit that derives an approximate straight line of the peak values ​​of the plurality of partial waveforms, and calculates the difference between the peak values ​​and the approximate straight line as the variation degree.

4. A tool life prediction device according to any one of claims 1 to 3, The remaining life prediction unit is a tool life prediction device that predicts the remaining life based on the difference between the degree of variation when the tool reaches the end of its life and the degree of variation calculated by the degree of variation calculation unit.

5. A tool life prediction device according to any one of claims 1 to 3, A tool life prediction device comprising a notification unit that notifies when the remaining life predicted by the remaining life prediction unit falls below a predetermined value.

6. A method for predicting tool life, A signal acquisition step involves acquiring an output signal from a sensor that detects a physical quantity generated when each of the multiple cutting edges of a tool having multiple cutting edges comes into contact with a workpiece being cut by the rotation of the tool, and acquiring an output signal from the sensor. A step of calculating the autocorrelation coefficient of the output signal, A variation degree calculation step for calculating the degree of variation among a plurality of partial waveforms that constitute the waveform of the autocorrelation coefficient, and which correspond to a plurality of blades, A remaining life prediction step that predicts the remaining life of the tool based on the degree of variation, A tool life prediction method comprising the following:

7. It is a machine tool, A spindle device having a spindle to which a tool having multiple blades is mounted, and which rotates the tool mounted on the spindle, The system includes a table on which a workpiece to be cut by the rotation of the tool is fixed, and a moving device that moves the workpiece fixed to the table relative to the tool mounted on the spindle, A control device that controls the spindle device and the moving device, A sensor for detecting a physical quantity that occurs when each of the multiple cutting edges comes into contact with the workpiece, Equipped with, The control device is A signal acquisition unit that acquires the output signal from the aforementioned sensor, An autocorrelation coefficient calculation unit that calculates the autocorrelation coefficient of the output signal, A variation degree calculation unit that calculates the degree of variation among a plurality of partial waveforms that constitute the waveform of the autocorrelation coefficient, and which corresponds to a plurality of blades, A remaining life prediction unit predicts the remaining life of the tool based on the degree of variation, A machine tool having

Images