MACHINE TOOL AND MACHINING SYSTEM

Abstract

Machine tool (10) for machining a workpiece, the machine tool comprising: a spindle (111) configured to rotate a holder mounted with a tool used for machining; one or more movable shafts (116) configured to move the holder (11) and / or a work surface on which the workpiece is arranged; a torque measuring unit (151) configured to measure the drive torque of the spindle (111) and / or the one or more movable shafts (116); a reference value calculation unit (152) configured to use, as a reference value, the drive torque measured by performing an idle run by rotating the spindle (111) in a normal state; a torque comparison unit (153) configured to compare the drive torque measured by performing an idle run by rotating the spindle (111) before the actual machining with the reference value; and an alarm unit (154) configured to determine whether to issue an alarm based on the results of the comparison.

Description

BACKGROUND OF THE INVENTIONField of the invention

[0001] The present invention relates to a machine tool. Related technology

[0002] In a machine tool, foreign objects such as chips can become trapped in the tool holder during tool replacement, for example. Machining errors such as workpiece misalignment, center misalignment, and uneven machining surfaces can occur when chips become trapped in the machine tool, and these errors lead to spindle overrun.

[0003] Fig. 1 is a diagram illustrating the clamping of foreign material 54 in a tapered portion 53 of a tool holder 52 holding a cutting tool 51. As in Fig. 1, the clamping of the foreign material 54 causes an eccentric displacement of the tool 51 as the spindle rotates.

[0004] In addition to the jamming of foreign material in the tool holder 52, the running of the cutting tool 51 held by the tool holder 52 can also be caused by incorrect assembly of the cutting tool 51, for example, by wear on the conical section, by a defective tool holder and by breakage or bending of the cutting tool 51.

[0005] A well-known technology for detecting cutting tool overrun in a machine tool is to install a sensor at the position where the cutting tool is held in the holder and use the sensor to directly detect cutting tool overrun (see, for example, Patent Documents 1 and 2).

[0006] Patent Document 1: Japanese Unexamined Patent Application Publication No. JP 2017 - 7 030 A.

[0007] Patent Document 2: Japanese Unexamined Patent Application Publication No. JP H08-197384 A.

[0008] DE 10 2017 005 349 A1 discloses a machine learning device, numerical control device, and machine learning method for learning a threshold value for detecting an abnormal load. A machine learning method for learning a threshold value for detecting an abnormal load in a machine tool includes a status observation unit and a learning unit. The status observation unit observes a status variable based on at least one of information about a tool, a main spindle rotation rate, and a coolant amount of the machine tool, a material of a workpiece, a movement direction, a cutting speed, and a cutting depth of the tool; and the learning unit learns the threshold value for detecting an abnormal load based on training data created from an output of the status observation unit and data related to detecting an abnormal load in the machine tool, and on the learning data. SUMMARY OF THE INVENTION

[0009] However, retrofitting a cutting tool coasting sensor presents numerous challenges, such as the high cost of each installed sensor, the wiring required for sensor installation, and the degradation of the sensor caused by cutting fluid. Therefore, there is a need for a machine tool capable of detecting cutting tool coasting without the need to add a cutting tool coasting sensor.

[0010] The problem is solved by a machine tool having the features of patent claim 1, by a machine tool having the features of patent claim 4 and by a machining system having the features of patent claim 8.

[0011] One aspect of the present disclosure is a machine tool for machining a workpiece, the machine tool comprising a spindle configured to rotate a holder mounted with a tool used for machining, one or more movable shafts configured to move the holder and / or a work surface on which the workpiece is placed, a torque measuring unit configured to measure the drive torque of the spindle and / or the movable shaft, a reference value calculating unit configured to use, as a reference value, the drive torque measured by performing an idle run by rotating the spindle in a normal state, a torque comparing unit configured to calculate the drive torque,which was measured by performing an idle run by rotating the spindle before the actual machining, with the reference value, and an alarm unit configured to determine whether an alarm should be issued based on the results of the comparison.

[0012] According to one aspect, a machine tool is capable of detecting the overrun of cutting tools without the need to add a sensor for detecting the overrun of cutting tools. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Fig. Figure 1 is a diagram illustrating the clamping of foreign material in a tapered section of a tool holder. Fig. 2 is a diagram illustrating a machine tool and a work surface on which a workpiece is placed according to an embodiment. Fig. 3 is a diagram illustrating a movement mechanism for a spindle in a machine tool according to an embodiment. Fig. Figure 4 is a graph illustrating the change of drive torque over time during normal operation. Fig. Figure 5 is a graph illustrating the time change of the drive torque during irregular operation. Fig. 6 is a functional block diagram of a machine tool according to an embodiment. Fig. 7 is a flowchart showing the operation of a machine tool according to an embodiment. Fig. 8 is a functional block diagram of a machine tool according to an embodiment. Fig. 9 is a flowchart illustrating the operation of a machine tool according to an embodiment. Fig. 10 is a graph showing the relationship between the spindle speed during idle and the reaction torque according to an embodiment. Fig. 11 is a graph showing the relationship between the spindle speed at idle and the reaction torque according to an embodiment. Fig. 12 is a diagram illustrating the entire configuration of a machining system according to an embodiment. Fig. 13 is a functional block diagram of a machine learning device incorporated into a processing system according to one embodiment. Fig. 14 is a flowchart illustrating the operation of a machine learning device included in a processing system according to one embodiment. DETAILED DESCRIPTION OF THE INVENTION <1. First Embodiment>

[0014] A first embodiment of the present invention will be described with reference to Fig. 2 to 7. Fig. 2 is a diagram illustrating a machine tool and a work surface on which a workpiece is arranged according to an embodiment. Fig. 3 is a diagram illustrating a moving mechanism for moving a spindle in the Fig. 2 in the direction of the Z-axis. Fig. Figure 4 is a graph illustrating the change of drive torque over time during normal operation. Fig. Figure 5 is a graph showing the time change of drive torque during irregular operation. Fig. 6 is a functional block diagram of the machine tool according to the present embodiment. Fig. 7 is a flowchart showing the operation of the machine tool according to the present embodiment. <1.1. Overview of the invention>

[0015] First, an overview of the invention according to the first embodiment will be given with reference to Fig. 2 to 5. In Fig. 2, a machine tool 10 includes a spindle and X, Y, and Z axes as motion axes for supporting the spindle. A cutting tool 12 is mounted on a holder 11 that rotates the spindle. The cutting tool 12 rotates to cut a workpiece 22, which is mounted on a jig 21 and placed on a work surface 20.

[0016] In Fig. 3, the holder 11 and the cutting tool 12 of the machine tool 10 include a rotating drive motor 13 for the movement axis, and this rotation is transmitted to a ball screw 15 via a toothed belt 14 to rotate the ball screw 15. The ball screw 15 is for linear movement along the Z-axis. As the ball screw 15 rotates, a nut 16 engaged with the ball screw 15 moves in the axial direction of the ball screw 15, causing the holder 11 fixed to the nut 16 to move in the Z-axis direction along the linear guide bearings 16A and 16B provided along the Z-axis direction. As a result, the spindle and the cutting tool 12 rotating around the spindle move in the Z-axis direction.

[0017] As above with reference to Fig. 1, occurs when the cutting tool 12 moves eccentrically because foreign material is clamped in the conical section of the holder 11, as shown in Fig. 2, eccentricity occurs when the spindle rotates and a reaction force is generated in the X, Y, and Z axes, the motion axes that support the spindle. This reaction force synchronizes with the rotation of the spindle and becomes the reaction torque of the drive motor 13 of the motion axis, as shown in Fig. 3. More precisely, if in Fig. 3 a reaction force synchronized with the rotation of the spindle is generated, this reaction force is transmitted to the ball screw 15 via the holder 11, which carries the cutting tool 12 rotating around the spindle, and further becomes the reaction torque of the drive motor 13 of the movement axis via the toothed belt 14.

[0018] As in Fig. 4, the drive torque fluctuates during normal operation at minute widths, but generally maintains a prescribed value. However, as described above, when eddies occur as a result of eccentricity synchronized with the spindle rotation, a reaction torque is generated, as shown in Fig. 5, and the drive torque has a periodic waveform synchronized with spindle rotation. Eccentricity whirl occurs when chips become trapped in the tapered section, the cutting tool moves eccentrically due to wear on the tapered section, the chuck is defective, or the cutting tool is broken or bent.

[0019] The reaction torque component, which includes the drive torque of the movable axis drive motor 13, is difficult to measure because this component is hidden in the drive torque. This is because a large force is generated by the reaction force of the machining itself.

[0020] Thus, the drive torque of the drive motor 13 of the movable axis is measured by rotating the spindle idly at a constant speed before machining is started, that is, when the cutting tool 12 is not in contact with the workpiece 22 and no machining is performed.

[0021] During normal operation, no reaction torque is generated due to the turbulence caused by eccentricity. Therefore, the drive torque on the moving axes at this time is taken as the normal reference. After setting the normal reference, the spindle rotates idle at a constant speed before starting actual machining, and the drive torque of the drive motor 13 of the moving axis is measured. This drive torque is then compared with the normal reference to detect any abnormality related to the overrun of the cutting tool 12 and to issue an alarm if an abnormality occurs.

[0022] The spindle of the machine tool 10 is often designed and assembled to rotate smoothly to ensure machining accuracy. Therefore, it is assumed that a disturbance component is smaller than a variance component of the drive torque during normal operation, that is, a periodic component associated with the vortex when eccentricity occurs.

[0023] In the present embodiment, assuming that the above-described fluctuation width (a1 in Fig. 5) of the drive torque generated during an idle run before the actual machining at a constant spindle speed can be compared with a reference value, which is a normal fluctuation range. Alternatively, the maximum value (a2 in Fig. 5) the drive torque is compared with a reference value, which is a maximum value during normal operation.

[0024] As a comparison method, the difference between a reference value as the fluctuation range of the drive torque generated during normal idling and a fluctuation range of the drive torque generated during idling before actual machining may be determined, and the absolute value of this difference may be compared with a threshold value corresponding to approximately 20% of the reference value. Alternatively, the difference between a reference value as the maximum value of the drive torque generated during normal idling and the maximum value of the drive torque generated during idling before actual machining may be determined, and the absolute value of this difference may be compared with a threshold value corresponding to approximately 20% of the reference value.Establishing a range for the fluctuation component caused by the disturbance makes it possible to reduce the false detection of anomalies. <1.2. Configuration of the invention>

[0025] Fig. 6 is a functional block diagram of the machine tool 10. As in Fig. 6, the machine tool 10 performs machining with a spindle 111 equipped with a cutting tool and rotated by a spindle motor 112, and a movable shaft 116 that extends the spindle 111. In other words, the spindle motor 112, which is configured to drive the spindle 111, rotates the cutting tool, and the cutting tool is extended by a movable shaft motor 117 that drives the movable shaft 116. In working examples, the machine tool 10 is described as a cutting machine, but the machine tool 10 is not limited thereto.

[0026] The machine tool 10 includes, in addition to the spindle 111, the spindle motor 112, the movable shaft 116, and the above-described movable shaft motor 117, a motor control circuit 130, motor-driven amplifiers 131A and 131B, and a control unit 150. The motor control circuit 130 calculates operation commands for driving the spindle motor 112 and the movable shaft motor 117. The motor-driven amplifier 131A amplifies the operation commands and outputs the operation commands to the spindle motor 112. The motor-driven amplifier 131B amplifies the operation commands and outputs the operation commands to the movable shaft motor 117.

[0027] The control unit 150 is a component that fully controls the machine tool 10. The control unit 150 reads and executes various programs from a memory area such as ROM, RAM, flash memory, or a hard disk drive (HDD) to realize the various functions of the present embodiment. The control unit 150 may be a CPU. The control unit 150 includes a torque measuring unit 151, a reference value calculating unit 152, a torque comparing unit 153, and an alarm unit 154.

[0028] The torque measuring unit 151 measures the drive torque of the spindle 111 and the movable shaft 116. The torque measuring unit 151 can calculate the drive torque based, for example, on the load current used to drive the spindle motor 112 and the movable shaft motor 117. Alternatively, the torque measuring unit 151 can measure the drive torque using magnetostrictive torque sensors mounted on the spindle 111 and the movable shaft 116.

[0029] The reference value calculation unit 152 uses, as a reference value, the drive torque measured by the torque measuring unit 151 when the machine tool 10 performs idle operation by rotating the spindle in a normal state. As described above, the reference value calculation unit 152 may use the maximum value of the normal drive torque or the overrun length of the drive torque as a reference value.

[0030] The torque comparison unit 153 compares the drive torque measured by idling the spindle of the machine tool 10 before the actual machining with the reference value. More specifically, the torque comparison unit 153 can compare the drive torque before the actual machining with the reference value by determining whether the absolute value of the difference between the drive torque measured before the actual machining and the reference value exceeds a threshold. The threshold can be a predetermined percentage of the reference value, for example, 20% of the reference value.

[0031] The alarm unit 154 determines whether to issue an alarm based on the comparison results of the torque comparison unit 153. Specifically, the alarm unit 154 may issue and alert when the torque comparison unit 153 determines that the absolute value of the difference between the drive torque before actual machining and the reference value has exceeded the threshold. <1.3. Operation of the invention>

[0032] Fig. 7 is a flowchart illustrating the operation of the machine tool 10. In step S11, the machine tool 10 performs an idle operation by rotating the spindle during normal operation based on the control of the motor control circuit 130.

[0033] In step S12, the torque measuring unit 151 measures the drive torque, and the reference value calculating unit 152 uses the measured value of the drive torque as a reference value.

[0034] In step S13, the machine tool 10 performs an idle operation by rotating the spindle before the actual machining based on the control by the motor control circuit 130.

[0035] In step S14, the torque measuring unit 151 measures the drive torque.

[0036] If the absolute value of the difference between the measured value and the reference value exceeds a threshold in step S15 (S15: YES), the processing proceeds to step S16. If the absolute value of the difference between the measured value and the reference value is equal to or less than the threshold (S15: NO), the processing ends.

[0037] In step S16, the alarm unit 154 issues an alarm. <1.4. Effects of the invention>

[0038] The machine tool 10 according to the present embodiment is a machine tool for machining a workpiece. The machine tool 10 includes the spindle 111 configured to rotate the holder 11 mounted with a tool used for machining, one or more movable shafts 116 configured to move the holder 11 and / or the work surface on which the workpiece 22 is placed, the torque measuring unit 151 configured to measure the drive torque of the spindle 111 and / or the movable shaft 116, a reference value calculating unit 152 configured to use, as a reference value, the drive torque measured by performing an idle run by rotating the spindle in a normal state, a torque comparing unit 153 configured to compare the drive torque,which was measured by performing an idle run by rotating the spindle before the actual machining, with the reference value, and an alarm unit 154 which is configured to determine whether an alarm should be issued based on the results of the comparison.

[0039] With this configuration, the machine tool is able to detect tool overrun without the need to add a tool overrun detection sensor.

[0040] Furthermore, the rotational speed of the spindle 111 during idling may be a constant value, the reference value calculation unit 152 may use the maximum value or the overrun length of the normal drive torque as a reference value, and the torque comparison unit 153 may compare the maximum value of the drive torque before machining or the overrun length with the reference value.

[0041] With this configuration, the torque comparison unit 153 can easily compare the drive torque and the reference value. <2. Second embodiment>

[0042] A second embodiment of the present invention will be described below with reference to Fig. 8 and Fig. 9 described. Fig. 8 is a functional block diagram of a machine tool according to the present embodiment. Fig. 9 is a flowchart illustrating the operation of the machine tool according to the present embodiment. <2.1. Sketch of the invention>

[0043] First, a sketch of the invention according to the second embodiment will be described.

[0044] In the first embodiment, the fluctuation width of the driving torque or the maximum value of the driving torque is compared with a normal reference value on the assumption that a disturbance component is smaller than a swirling periodic component when eccentricity occurs.

[0045] Therefore, a frequency analysis of the torque generated on the movable shaft is performed to extract only a specific frequency component synchronized with the spindle speed, and the value of this frequency component is compared with the normal reference value. As a result, it is possible to separate a disturbance torque component in the drive torque generated by an element such as a bearing in the drive system, and therefore, it is not necessary to assume that the disturbance component is smaller than the whirling periodic component when eccentricity occurs.

[0046] The value of the frequency component can be, for example, an amplitude value. The amplitude value can be peak amplitude, root mean square (RMS) amplitude, power, or power spectral density (PSD). By comparing the measured value with the normal reference value before the start of actual machining and using the amplitude value as the value of the frequency component, it is possible to detect the eccentricity caused by the machine tool overrun during spindle rotation. <2.2. Configuration of the invention>

[0047] Fig. 8 is a functional block diagram of a machine tool 10A. In the following description, only the differences between the machine tool 10A and the machine tool 10 are primarily described, and descriptions of common components are generally omitted.

[0048] Machine tool 10A differs from machine tool 10 in that machine tool 10A includes a control unit 150A instead of control unit 150A. Control unit 150A differs from control unit 150A in that control unit 150A does not include torque comparison unit 153 and instead includes a frequency component extraction unit 155 and a frequency component comparison unit 156. Control unit 150A also includes a reference value calculation unit 152A instead of reference value calculation unit 152 and an alarm unit 154A instead of alarm unit 154.

[0049] The frequency component extraction unit 155 performs fast Fourier transform (FFT) analysis of the drive torque measured by the torque measuring unit 151 to thereby extract a frequency component synchronized with the rotational speed of the spindle 111. Specifically, the frequency component extraction unit 155 performs FFT calculation of the drive torque of the spindle 111 and / or the movable shaft 116 measured by the torque measuring unit 151 to obtain the power spectrum of the drive torque, and outputs the value of the frequency component synchronized with the rotational speed of the spindle 111 to the reference value calculation unit 152A and the frequency component comparison unit 156, which will be described later.For example, as described above, the value of the frequency component can be an amplitude value, and the amplitude value can be peak amplitude, root mean square (RMS), power, or power spectral density (PSD).

[0050] The reference value calculation unit 152A uses the value of the frequency component extracted by the frequency component extraction unit 155 when the machine tool 10A performs idling by rotating the spindle in a normal state as a reference value.

[0051] The frequency component comparison unit 156 compares the reference value and the value of the frequency component extracted by the frequency component extraction unit 155 by the machine tool 10A, performing an idle run by rotating the spindle before the actual machining. Specifically, the frequency component comparison unit 156 may compare the reference value and the value of the frequency component before the actual machining by determining whether the absolute value of the difference between the reference value and the value of the frequency component extracted before the actual machining exceeds a threshold. In this case, the threshold may be a predetermined percentage of the reference value, for example, 20% of the reference value.

[0052] The alarm unit 154A determines whether to issue an alarm based on the comparison results of the frequency component comparison unit 156. Specifically, the alarm unit 154A may issue an alarm when the frequency component comparison unit 156 determines that the absolute value of the difference between the reference value and the value of the frequency component before the actual processing exceeds the threshold. <2.3. Operation of the invention>

[0053] Fig. 9 is a flowchart illustrating the operation of the machine tool 10A. In step S21, the machine tool 10A performs an idle operation by rotating the spindle during normal operation based on the control by the motor control circuit 130.

[0054] In step S22, the torque measuring unit 151 measures the driving torque. In step S23, the frequency component extraction unit 155 extracts the frequency component from the measured driving torque, and the reference value calculation unit 152A uses the value of the extracted frequency component as a reference value.

[0055] In step S24, the machine tool 10A performs an idle operation by rotating the spindle before the actual machining based on the control by the motor control circuit 130.

[0056] In step S25, the torque measuring unit 151 measures the drive torque.

[0057] In step S26, the frequency component extraction unit 155 extracts the frequency component from the measured drive torque and uses the value of the extracted frequency component as a measured value.

[0058] If the absolute value of the difference between the measured value and the reference value exceeds the threshold in step S27 (S27: YES), the processing proceeds to step S28. If the absolute value of the difference between the measured value and the reference value is equal to or less than the threshold (S27: NO), the processing ends.

[0059] In step S28, the alarm unit 154A issues an alarm. <2.4. Effects of the invention>

[0060] The machine tool 10A according to the present embodiment is a machine tool for machining a workpiece, the machine tool including the spindle 111 configured to rotate the holder 11 mounted with a tool used for machining, one or more movable shafts 116 configured to move the holder 11 and / or a work surface on which the workpiece 22 is placed, the torque measuring unit 151 configured to measure the drive torque of the spindle 111 and / or the movable shaft 116, the frequency component extraction unit 155 configured to analyze the drive torque using FFT and extract a frequency component synchronized with the rotational speed of the spindle 111, the reference value calculation unit 152A configured to use a value of the frequency component as a reference value,extracted by performing an idle run by rotating the spindle in a normal state, the frequency component comparison unit 156 is configured to compare the value of the frequency component extracted by performing an idle run by rotating the spindle before the actual machining with the reference value, and the alarm unit 154A is configured to determine whether to issue an alarm based on the results of the comparison.

[0061] Even if this configuration does not assume that the disturbance component is smaller than the swirling periodic component when the eccentricity occurs, this configuration can detect the tool lag without adding a tool lag detection sensor. <3. Third Embodiment>

[0062] A third embodiment of the present invention will be described below with reference to Fig. 10 described. Fig. 10 is a graph showing the relationship between the value of the reaction torque and the rotational speed of the spindle 111 during idling according to the present embodiment. <3.1. Sketch of the invention>

[0063] First, a sketch of the invention according to the third embodiment will be described. There may also be a case where machining is affected because the eccentricity and reaction torque are small, but it is difficult to detect the reaction torque because the reaction force is obscured by the disturbance. In such a case, the reaction torque at a small eccentricity can be emphasized by using a resonance point based on the rigidity in the mechanical system supporting the spindle 111 and a resonance phenomenon using the rotation of the eccentricity as excitation.

[0064] More precisely, in Fig. 10, when the spindle speed during idling is aligned with the resonance point of the machine system, the value of a speed-periodic component in the reaction torque reaches a peak value. This value can be used to emphasize the reaction torque by using the spindle speed during idling as the resonance point where resonance occurs. <3.2. Configuration and operation of the invention>

[0065] The configuration itself of the machine tool according to the third embodiment is generally the same as the machine tool 10 according to the first embodiment and the machine tool 10A according to the second embodiment, and therefore, a description thereof will be omitted.

[0066] In the first and second embodiments, the rotational speed of the spindle motor 112 and / or the movable shaft motor 117 controlled by the motor control circuit 130 is a constant value; in the third embodiment, the rotational speed is a resonance point at which resonance occurs.

[0067] By using the spindle speed as the resonance point, the resonance point can be found in advance and the spindle speed can be aligned with the resonance point during idling. Furthermore, the speed can fluctuate between a low speed and a high speed that should encompass the resonance point, and the value measured in this section can be compared with a threshold value based on a3 in Fig. 10 to match the resonance point. <3.3. Effects of the invention>

[0068] In the present embodiment, the rotational speed of the spindle 111 during idling is the rotational speed at which resonance occurs in the machine tool 10 or 10A.

[0069] With this configuration, even when the reaction force is obscured by noise and the reaction torque is difficult to detect, tool overrun can be detected without adding a tool overrun detection sensor. <4.1. Sketch of the invention>

[0070] First, a sketch of the invention according to a fourth embodiment will be described. In the third embodiment, a resonance point is basically detected in advance, and then a spindle speed during idling is adjusted to match the resonance point. On the other hand, a resonance point may be changed by a rigidity difference between machine systems or by a tool type connected to a machine system. In such a case, it may be difficult to detect a resonance point in advance.

[0071] In this context, as in Fig. As shown in Figure 11, while a spindle speed is increased from a low speed to a high speed during idling, the value of a speed-periodic component in the reaction torque is compared with a threshold value larger than the value of a speed-periodic component in the normal state, and a point at which the threshold value is exceeded is estimated as a peak value. Then, a reference value and a drive torque at a speed at the time when the value of the speed-periodic component in the reaction torque reaches the peak value are compared, and based on the comparison result, it is determined whether to issue an alarm. <4.2. Configuration and operation of the invention>

[0072] The configuration itself of a machine tool according to the fourth embodiment is generally the same as the machine tool 10 according to the first embodiment and the machine tool 10A according to the second embodiment, and therefore, a description thereof will be omitted.

[0073] In the first and second embodiments, the rotational speed of the spindle motor 112 and / or the movable shaft motor 117 controlled by the motor control circuit 130 is a constant value. In the third embodiment, this rotational speed is a resonance point where resonance occurs. In the fourth embodiment, the rotational speed of the spindle motor 112 and / or the movable shaft motor 117 controlled by the motor control circuit 130 is fluctuated from a low rotational speed to a high rotational speed, and a reference value and a drive torque at a rotational speed at the time point at which the value of a periodic component of the rotational speed in the reaction torque is estimated to have reached a peak are compared.

[0074] If there are two or more points each estimated as a peak, a point with the highest value of the periodic component of the speed in the reaction torque is estimated as a peak.

[0075] After determining a speed at the time when the value of the periodic component of the speed is estimated to have reached a peak, the determined speed can be stored for each tool type in a memory unit (not shown). This allows a second and subsequent measurements to measure a reference value and drive torque using a spindle speed fixed from the first to a speed associated with a tool. <4.3. Effects of the invention>

[0076] In the present embodiment, although the rotational speed of the spindle 111 is increased from a low speed to a high speed during idling, a rotational speed at which resonance occurs is estimated.

[0077] With this configuration, even if a resonance point is not known in advance, tool overrun can be detected without adding a tool overrun detection sensor. <5. Fifth Embodiment>

[0078] A fifth embodiment of the present embodiment will be described with reference to the Fig. 12 to 14 described. Fig. 12 is a diagram illustrating the entire configuration of a machining system 1 according to the present embodiment. Fig. 13 is primarily a functional block diagram of a machine learning device 70 comprising the processing system 1. Fig. 14 is a flowchart illustrating the operation of the machine learning device 70. <5.1. Sketch of the invention>

[0079] In the first to fourth embodiments, the drive torque data itself or the value of the frequency component extracted from the drive torque data is compared between normal operation and irregular operation, and an alarm is issued when irregular operation occurs. In the fifth embodiment, machine learning is employed using learning data. This learning data is composed of data in which the drive torque during normal operation and a label indicating that the cutting tool 12 is correctly mounted in the machine tool 10 form a pair, and data in which the drive torque when an abnormality is intentionally generated and a label indicating that the cutting tool 12 is incorrectly mounted in the machine tool 10 form a pair. That is, "supervised learning" is performed.A learning model is created through supervised learning, and this learning model is used to determine the mounting state of the cutting tool 12 and to decide whether to issue an alarm based on the results of the determination. <5.2. Configuration of the invention>

[0080] As in Fig. 12, the machining system 1 includes a machine learning device 70 and a number of n machine tools 10 (where n is a natural number).

[0081] The machine learning device 70 and the machine tools 10 are communicatively connected to each other. The machine learning device 70 and the machine tools 10 can be connected directly via a connection interface or via a network 40. The network 40 can be a local area network (LAN) installed in a factory, the Internet, a public switched telephone network, or a combination of these networks. The specific communication method in the network 40 can be wired communication or wireless communication and is not particularly limited.

[0082] The machine learning device 70 uses supervised machine learning to create a learning model for detecting an abnormal installation state of the cutting tool 12 in the machine tool 10. To achieve this, the machine learning device 70 comprises, as shown in Fig. 12, an input unit 71, a label acquisition unit 72, a learning unit 73 and a learning model storage unit 74.

[0083] The input unit 71 acquires data relating to the drive torque as a feature value from the machine tool 10. The label acquisition unit 72 acquires labels indicating that the cutting tool 12 is correctly or incorrectly mounted in the machine tool 10.

[0084] The learning unit 73 performs supervised learning by using pairs of feature values and labels as learning data to create a learning model that can determine whether the cutting tool 12 is correctly or incorrectly mounted in the machine tool 10. The created learning model is sent to the control unit 150 of the machine tool 10. The learning model storage unit 74 stores the learning model created by the learning unit 73.

[0085] The learning unit 73 may use a support vector machine (hereinafter referred to as "SVM") as an example. An SVM is a well-known technology and will not be described in detail here. SVM is an identification technique that uses supervised learning (learning in which a computer receives correct data and incorrect data as learning data) and is known as a learning model with high identification accuracy. For example, an SVM is known to achieve high identification accuracy (high generalization ability) with unlearned data.

[0086] The learning unit 73 uses binarized labels related to the mounted state of the cutting tool 12 for the labels described above, and calculates a hyperplane that divides the space including the feature value described above such that the edges have a maximum with respect to whether the cutting tool 12 is correctly or incorrectly mounted. Furthermore, the learning unit 73 may use a coefficient of the hyperplane as a parameter of the learning model used by the machine tool 10 to determine the mounted state of the cutting tool 12.

[0087] As described above, the learning model storage unit 74 stores the learning model created by the learning unit 73. <5.3. Operation of the invention>

[0088] Fig.14 is a flowchart illustrating the operation of the machine learning device 70 when performing machine learning. In step S31, the input unit 71 of the machine learning device 70 acquires the feature values from the machine tool 10. More specifically, the input unit 71 of the machine learning device 70 acquires drive torque data when the cutting tool is correctly mounted and drive torque data when the cutting tool is incorrectly mounted.

[0089] In step S32, the label detection unit 72 of the machine learning device 70 detects labels indicating whether the cutting tool is correctly or incorrectly mounted.

[0090] In step S33, the learning unit 73 of the machine learning device 70 receives a pair of a feature value and a label as learning data.

[0091] In step S34, the learning unit 73 of the machine learning device 70 performs machine learning using the learning data.

[0092] In step S35, the learning unit 73 of the machine learning device 70 determines whether to terminate machine learning or repeat it. The condition for terminating machine learning can be arbitrarily set. For example, machine learning can be terminated when machine learning has been performed a predetermined number of times.

[0093] If the machine learning is to be repeated, processing returns to step S31, and the machine learning device 70 performs the same operation. If the machine learning is to be terminated, the machine learning device 70 sends the learning model created up to that point by machine learning to each machine tool 10 via the network 40 in step S36.

[0094] The learning model storage unit 74 of the machine learning device 70 stores the learning model. With this configuration, when a newly installed machine tool 10 requests the learning model, the learning model can be sent to that machine tool 10. When new learning data is acquired, additional machine learning can be performed on the learning model. <5.4. Effects of the invention>

[0095] In the present embodiment, the learning model created by machine learning can be shared by a plurality of machine tools 10. <6. Other embodiments>

[0096] The above-described embodiments are preferred embodiments of the present invention. However, these embodiments are not intended to limit the scope of the present invention, and the present invention can be implemented in various ways without departing from the gist of the present invention. <6.1. Modification Example 1>

[0097] The machining system according to the fourth embodiment includes the machine learning device 70 and the number of n machine tools 10 according to the first embodiment, but the machining system is not limited thereto. For example, instead of the number of n machine tools 10, the machining system may include the machine tool 10A according to the second embodiment or the machine tool 10 or 10A according to the third embodiment.

[0098] When using the machine tool 10A, the input unit 71 of the machine learning device 70 acquires the value of the frequency component instead of the drive torque data as the feature value from the machine tool 10.

[0099] Furthermore, the machine learning device 70 may be integrated into the machine tool 10 or 10A to integrate the machine tool 10 or 10A and the machine learning device 70. <6.2. Modification Example 2>

[0100] In the first embodiment, the torque measuring unit 151 measures the driving torque of the movable shaft 116, but the torque measuring unit 151 is not limited to this. For example, the torque measuring unit 151 may measure the holding torque instead of the movable shaft 16 when each shaft is in a stationary state. The holding torque is obtained by adding the eddy torque to the static friction torque. <6.3. Modification Example 3>

[0101] In Modification Example 2, there may be a case where the reaction force is measured while the movable shaft is stationary, but the reaction force may be difficult to measure due to a static friction torque component in the movable shaft drive system. Consequently, in Modification Example 3, the torque measuring unit 151 may use the drive torque during low-speed movement of, for example, 100 mm / s instead of the drive torque of the movable shaft 116, or the holding torque when the shafts are stationary. The drive torque in this case is obtained by adding the swirl torque to the sliding friction torque instead of the static friction torque. In some cases, it is easier to detect the reaction force component when the sliding friction component becomes dominant due to the slow movement of the moving shaft. REFERENCE SYMBOL 1 processing system 10, 10A machine tool 40 Network 70 Machine learning device 71 Input unit 72 Label capture unit 73 learning units 74 Learning model storage unit 111 spindle 113 spindle motor 116 Movable shaft 117 Motor with movable shaft 130 Motor control circuit 131A, 131B Motor drive amplifier 150, 150A control unit 151 Torque measuring unit 152, 152A Reference value calculation unit 153 Torque comparison unit 154, 154A alarm unit 155 Frequency component extraction unit 156 Frequency component comparison unit

Claims

[1] Machine tool (10) for machining a workpiece, the machine tool comprising: a spindle (111) configured to rotate a holder mounted with a tool used for machining; one or more movable shafts (116) configured to move the holder (11) and / or a work surface on which the workpiece is arranged; a torque measuring unit (151) configured to measure the drive torque of the spindle (111) and / or the one or more movable shafts (116); a reference value calculation unit (152) configured to use, as a reference value, the drive torque measured by performing an idle run by rotating the spindle (111) in a normal state; a torque comparison unit (153) configured to compare the drive torque measured by performing an idle run by rotating the spindle (111) before the actual machining with the reference value; and an alarm unit (154) configured to determine whether to issue an alarm based on the results of the comparison. [2] Machine tool (10) according to claim 1, wherein: the speed of the spindle (111) is a constant value during idle running; the reference value calculation unit (152) uses a maximum value or a lag length of the drive torque during normal operation as a reference value; and the torque comparison unit (153) compares the maximum value of the drive torque or the overrun length before machining with the reference value. [3] Machine tool (10) according to claim 1, wherein: the speed of the spindle (111) is a constant value during idle running; and the torque measuring unit (151) measures the holding torque when each movable shaft (116) is in a stationary state, or the driving torque when each movable shaft (116) moves at a speed equal to or lower than a predetermined speed. [4] Machine tool (10A) for machining a workpiece, the machine tool comprising: a spindle (111) configured to rotate a holder mounted with a tool used for machining; one or more movable shafts (116) configured to move the holder and / or a work surface on which the workpiece is arranged; a torque measuring unit (151) configured to measure the drive torque of the spindle (111) and / or the one or more movable shafts (116); a frequency component extraction unit (155) configured to analyze the drive torque by FFT and extract a frequency component synchronized with the rotational speed of the spindle (111); a reference value calculation unit (152A) configured to use, as a reference value, a value of the frequency component extracted by performing an idle run by rotating the spindle (111) in a normal state; a frequency component comparison unit (156) configured to compare the value of the frequency component extracted by performing an idle run by rotating the spindle (111) before the actual machining with the reference value; and an alarm unit (154A) configured to determine whether to issue an alarm based on the results of the comparison. [5] The machine tool (10A) according to any one of claims 1 to 4, wherein the rotational speed of the spindle (111) during idling is a rotational speed at which resonance occurs in the machine tool (10A). [6] The machine tool (10A) according to any one of claims 1 to 5, wherein the rotational speed of the spindle (111) during idling is a rotational speed at the time point at which the value of a periodic rotational speed component in the reaction torque of the spindle (111) is estimated to be a peak value while the rotational speed of the spindle (111) fluctuates from a low rotational speed to a high rotational speed. [7] The machine tool (10A) according to claim 6, further comprising a storage unit that stores the rotational speed at the time when the value of the periodic component of the rotational speed in the reaction torque of the spindle (111) is estimated to have become the peak value for each type of the tool, the rotational speed of the spindle (111) during idling being the rotational speed stored in the storage unit. [8] Processing system (1), comprising: one or more of the machine tools (10, 10A) according to one of claims 1 to 7; and a machine learning device (70) comprising: a label detection unit (72) configured to detect data indicating whether the tool is correctly or incorrectly attached in the machine tool (10, 10A) as labels; and a learning unit (73) configured to create a learning model for identifying a mounted state of the tool by performing supervised learning, using as learning data the drive torque measured when the tool is correctly mounted and a label indicating that the tool is correctly mounted as a pair, and the drive torque measured when the tool is incorrectly mounted and a label indicating that the tool is incorrectly mounted as a pair. [9] The machining system (1) according to claim 8, wherein the machine learning device (70) is built into the machine tool (10) to integrate the machine learning device (70) and the machine tool (10).

Images