JPWO2019186620A1 - Abnormality sign detector and machine tool - Google Patents

Abstract

The abnormality sign detection device that detects the sign of abnormality of the tool used for machining is collected by the data collection unit that collects the vibration data of the tool during machining and the data collection unit during the predetermined period at the beginning of machining. Whether or not the vibration data is included in the normal region is determined by using the model construction unit that builds the judgment model in the one-class support vector machine using the vibration data as the data in the normal region and the judgment model constructed by the model construction unit. It is provided with a sign detection unit that detects the presence or absence of a sign of abnormality of the tool by determining.

Description

The present specification discloses an abnormality sign detection device and a machine tool.

Conventionally, it has been known that data on the state of a tool is acquired during machining using a tool (cutting body) such as a cutting tool, and an abnormality of the tool is detected based on the data. For example, in Patent Document 1, the vibration of the tool is sampled at a fixed cycle, and the parameters of the model are calculated based on the current sampling value and the past sampling values of a plurality of times. Then, the degree of deviation between the reference value of the parameter and the calculated current parameter is obtained, and when the obtained degree of deviation reaches a predetermined management standard, an abnormality of the tool is detected.

Japanese Unexamined Patent Publication No. 57-54053

Detecting such abnormalities in tools can be said to be an important issue in machine tools that perform machining using tools. However, since the above-mentioned method requires sampling a plurality of times, there is still room for improvement in order to quickly detect a sign of an abnormality in the tool.

A main object of the present disclosure is to enable proper use of a tool by making it possible to detect a sign of an abnormality of the tool at an early stage.

The present disclosure has taken the following steps to achieve the above-mentioned main objectives.

The abnormality sign detection device of the present disclosure is an abnormality sign detection device that detects a sign of abnormality of a tool used for machining, and includes a data collecting unit that collects vibration data of the tool during machining and the machine. A model construction unit that builds a judgment model in a one-class support vector machine using the vibration data collected by the data collection unit as data in a normal region during a predetermined period at the beginning of processing, and the model construction unit constructed by the model construction unit. The gist is to include a sign detection unit for detecting the presence or absence of a sign of abnormality of the tool by determining whether or not the vibration data is included in the normal region using a determination model.

The abnormality sign detection device of the present disclosure constructs a judgment model in a one-class support vector machine using vibration data collected in a predetermined period at the beginning of machining as data in a normal region, and the vibration data is in the normal region of the judgment model. By determining whether or not it is included, the presence or absence of a sign of an abnormality in the tool is detected. As a result, the determination model can be constructed in a predetermined period at the beginning of machining, and the detection of signs of abnormality can be started promptly. Further, for the same type of tool that performs machining with the same content, it is possible to detect a sign of abnormality from the beginning of machining by using the already constructed determination model. Therefore, it is possible to detect a sign of an abnormality of the tool at an early stage and enable proper use of the tool.

The block diagram which shows the schematic example of the structure of the machine tool 10. Explanatory drawing which shows the electrical connection relation of the machine tool 10. The functional block diagram of the control device 50 concerning the detection of a sign of abnormality. Explanatory drawing which shows an example of time transition of vibration data. The flowchart which shows an example of the abnormality sign detection processing. Explanatory drawing which shows an example of threshold value Th and similarity Ds. Explanatory drawing which shows an example of time transition of vibration data and similarity Ds. Explanatory drawing which shows change of similarity Ds with model reconstruction.

Next, an embodiment of the present disclosure will be described with reference to the drawings. FIG. 1 is a configuration diagram showing a schematic example of the configuration of the machine tool 10, and FIG. 2 is an explanatory diagram showing an electrical connection relationship of the machine tool 10. The left-right direction in FIG. 1 is the Y-axis direction, the up-down direction is the Z-axis direction, and the direction perpendicular to the paper surface in FIG. 1 is the X-axis direction.

The machine tool 10 is configured as a machining device such as a machining center that performs machining work on a work W using a tool T such as a tip or a drill. As shown in FIGS. 1 and 2, the machine tool 10 includes a base 12 as a gantry, a table 14 and a column 16 provided on the base 12, a work chuck device 20, a spindle device 30, and a spindle moving mechanism. 40 and a control device 50 are provided. Further, the machine tool 10 includes an operation panel 18 and a notification device 19. The operation panel 18 is configured as a touch panel type liquid crystal screen, and displays various instruction screens such as processing menus, change screens of various settings, status screens showing work status, and selection operations and setting operations by the operator. Or accept. In addition, the operation panel 18 includes a power button for turning on / off the power of the machine tool 10, a start button for starting the operation of the machine tool 10, a stop button for stopping the operation, and the like. The alarm 19 is composed of a speaker that outputs a warning sound, a voice indicating an abnormality, and a rotating light that lights up in a mode for notifying an abnormality. The machine tool 10 is provided with an automatic tool changing device (not shown), and the tool T mounted on the spindle device 30 can be automatically changed according to the content of machining.

The work chuck device 20 is composed of a mounting table 22 arranged on the table 14 on which the work W is mounted, a servomotor and a ball screw mechanism (not shown), and the like, and a Y-axis moving mechanism for moving the mounting table 22 in the Y-axis direction. 24, a plurality of pairs of chuck claws 26 for chucking the work W mounted on the mounting table 22, and an opening / closing device (not shown) for opening / closing the chuck claws 26 are provided. The spindle device 30 includes a spindle 32 to which a tool T is replaceably mounted on the tip thereof, a spindle motor 34 for rotationally driving the spindle 32, and a vibration sensor 36 for detecting vibration of the spindle 32. The spindle moving mechanism 40 is composed of a servomotor and a ball screw mechanism (not shown), respectively, and has a Z-axis moving mechanism 42 for moving the spindle device 30 in the Z-axis direction and a Z-axis moving mechanism provided on the column 16 for moving the spindle device 30. It is provided with an X-axis moving mechanism 44 that moves the entire 42 in the X-axis direction.

The control device 50 includes a CPU, ROM, RAM, various interfaces, etc. (not shown). The detection value of the vibration sensor 36 of the spindle device 30, the position signal from each servomotor, the operation signal received by the operation panel 18, and the like are input to the control device 50. Further, the control device 50 includes a servomotor and an opening / closing device of the Y-axis moving mechanism 24 of the work chuck device 20, a spindle motor 34 of the spindle device 30, a Z-axis moving mechanism 42 of the spindle moving mechanism 40, and an X-axis moving mechanism 44. A drive signal is output to the servomotor, a display signal is output to the operation panel 18, and a notification command of an abnormality sign alarm indicating a sign of abnormality is output to the alarm device 19.

Further, the control device 50 has a function of detecting a sign of abnormality of the tool T by a well-known 1-class support vector machine (hereinafter, 1-class SVM) using the vibration data from the vibration sensor 36. FIG. 3 is a functional block diagram of the control device 50 for detecting a sign of abnormality. As shown in the figure, the control device 50 includes a data collection unit 51, a feature amount extraction unit 53, a model construction unit 54, a similarity calculation unit 56, a threshold value change unit 57, and an abnormality sign notification unit 58. ..

The data collection unit 51 collects the detection voltage (V) of the vibration sensor 36 of the spindle device 30 as vibration data and stores it in the vibration data area 52 such as a RAM. FIG. 4 is an explanatory diagram showing an example of time transition of vibration data, in which the horizontal axis represents time (seconds) and the vertical axis represents the detection voltage (V) of the vibration sensor 36. FIG. 4 shows an example of vibration data during a plurality of machining operations of the machine tool 10, for example, a plurality of drilling operations on one or a plurality of workpieces W. In the machine tool 10, while the tool T such as a drill for drilling is continuously used, the tool T is worn and fatigue is accumulated in the tool T, so that the vibration in the normal state at the beginning of machining occurs. The difference between the two becomes large, and the tool T may break depending on the degree of accumulation of fatigue. In the example of FIG. 4, the vibration data at the right end is remarkably large, and shows the case where the tool T is broken. It is desirable for the operator to replace the tool T at an appropriate timing before the breakage of the tool T occurs, but it is difficult for the operator to determine an appropriate replacement time for the tool T. Therefore, although the tool T can still be sufficiently used, the operator may replace the tool T early, or the replacement of the tool T by the worker may be delayed and breakage may occur. In particular, when a used tool T is used in the machine tool 10, it is more difficult for the operator to determine the replacement time because the degree of accumulation of fatigue of the tool T is unknown or individual differences are large for each tool T. Become.

The feature amount extraction unit 53 extracts the frequency feature amount by performing, for example, a fast Fourier transform (FFT) on the vibration data collected by the data collection unit 51. The model building unit 54 treats the feature amount of the frequency extracted by the feature amount extracting unit 53 in a predetermined learning period as data in the normal region, and constructs a determination model 55 used for abnormality determination by the one-class SVM. The similarity calculation unit 56 includes a threshold value Th as a determination standard for classifying the data in the normal region and the other data in the determination model 55 of the 1-class SVM, and the feature amount of the frequency extracted by the feature amount extraction unit 53. Are compared, and the similarity Ds indicating the distance from the threshold value Th is calculated. In the abnormality determination in the 1st class SVM, when the similarity Ds calculated by the similarity calculation unit 56 is 0 or more, it is determined to be normal, and when the similarity Ds is less than 0, it is determined to be abnormal. The threshold value changing unit 57 changes the threshold value Th setting based on the threshold value Th setting change operation received via the operation panel 18. When the similarity Ds calculated by the similarity calculation unit 56 is less than 0, the abnormality sign notification unit 58 outputs a notification command to the notification device 19 so as to notify the fact that the abnormality is a sign.

Next, the operation of the machine tool 10 of the present embodiment configured in this way will be described. Here, the operation of detecting a sign of abnormality will be described. FIG. 5 is a flowchart showing an example of abnormality sign detection processing. This process is executed by each function related to the above-mentioned sign detection of abnormality of the control device 50 in a state where the machine tool 10 is ready for machining. When this process is started, the control device 50 first waits for the start of machining using the tool T (S100), and when the machining is started, the control device 50 is based on the vibration data collected from the vibration sensor 36. A determination model 55 of 1-class SVM is constructed (S110). The construction of the determination model 55 is performed until the predetermined learning period ends (S120). The predetermined learning period can be, for example, a period from the start of the drilling process by the machine tool 10 to the end of the drilling process several times. In this way, the control device 50 builds the determination model 55 based on the vibration data at the beginning of machining. This is because the tool T is unlikely to break during the initial period when machining is started, and the vibration data obtained during that period can be handled as data in a normal state.

When the predetermined learning period ends, the control device 50 determines whether or not the threshold value Th setting change operation is accepted via the operation panel 18 (S130) and whether or not machining is in progress (S150). To do. When the control device 50 determines that the setting change operation has been accepted in S130, the control device 50 changes the threshold Th setting based on the setting change operation (S140). Although not shown here, the operator can adjust the threshold value Th to the positive side or the negative side via the setting screen of the operation panel 18.

Further, when the control device 50 determines that machining is being performed in S150, whether or not the vibration data collected from the vibration sensor 36 using the determination model 55 constructed in S110 is included in the normal region of the determination model 55. The sign of abnormality is detected based on (S160). Further, the control device 50 determines whether or not there is a sign of abnormality in the processing of S160 (S170) and whether or not the ongoing machining (one processing) is completed (S180), respectively. The processing of S160 is continued. In S160, as described above, the control device 50 compares the feature amount of the frequency extracted from the vibration data with the threshold value Th, calculates the similarity Ds indicating the distance between the feature amount and the threshold value Th, and calculates the similarity degree Ds. When the value of Ds is less than 0, it is determined in S170 that there is a sign of abnormality. When the similarity Ds is 0 or more, the control device 50 determines in S170 that there is no sign of abnormality.

Here, FIG. 6 is an explanatory diagram showing an example of the threshold value Th and the similarity Ds. In FIG. 6, the circles indicate the data of the determination model 55 collected during the learning period, and the □ and Δ indicate the data after the model is constructed. Further, the dotted line indicates the threshold value Th in the normal region, and the control device 50 determines that there is no sign of abnormality if the similarity Ds is positive □, and the similarity Ds is negative △. If there is, it is judged that there is a sign of abnormality. When the threshold value Th is changed to the negative side by the threshold value changing unit 57, the similarity Ds calculated by the similarity calculation unit 56 tends to be a value of 0 or more, so that it is easy to determine that it is normal. Become. On the other hand, when the threshold value Th is changed to the positive side by the threshold value changing unit 57 as compared to before the change, the similarity Ds calculated by the similarity calculation unit 56 tends to be less than 0, so that it is easy to determine that it is abnormal. Become.

When the control device 50 determines that the current machining has been completed (S180) while determining that there is no sign of abnormality in S170, it determines whether or not there is the next machining (S190), and the next machine. If it is determined that there is no processing, the abnormality sign detection process is terminated. On the other hand, when the control device 50 determines in S190 that there is the next machining, it determines whether or not the machining contents (machining conditions) in the next machining are the same (S200). In S200, the control device 50 is based on whether or not the material, shape, size, machining position, machining allowance, type of tool T used in machining, and the like of the work W to be machined are the same. , Judge whether the processing contents are the same. When the control device 50 determines that the processing contents are the same, the control device 50 returns to S150 and repeats the processing.

In this case, when the next machining is started, the control device 50 will detect a sign of abnormality using the determination model 55 already constructed. That is, once the determination model 55 is constructed, the control device 50 continues to use the same determination model 55 for machining with the same machining content without constructing a new determination model 55. Therefore, for example, when a used tool T is used, if the determination model 55 of the same type of tool T has already been constructed, it is possible to detect a sign of abnormality from the beginning of machining using the used tool T. It will be possible. As described above, since it is difficult for the operator to determine the replacement time of the used tool T, it is possible to detect a sign of abnormality from the beginning of machining to prevent the used tool T from breaking. It becomes possible to use it properly. Further, when the control device 50 determines in S200 that the processing contents of the next machining are not the same, the control device 50 returns to S100 and performs the processing. That is, when the processing content is switched, the control device 50 newly constructs the determination model 55.

Further, when the similarity Ds becomes less than 0, the control device 50 determines in S170 that there is a sign of abnormality, and outputs an abnormality sign alarm indicating the sign of abnormality from the alarm 19 (S210). The control device 50 may display a warning message indicating a sign of abnormality on the operation panel 18.

FIG. 7 is an explanatory diagram showing an example of the time transition of the vibration data and the similarity Ds. As described above, the determination model 55 is constructed in a predetermined learning period after the start of machining. Further, the similarity Ds gradually approaches the value 0 due to the change in the vibration data due to the change in the state such as the wear of the tool T and the accumulation of fatigue. Then, the control device 50 determines that there is a sign of abnormality at the time T1 when the similarity Ds falls below the value 0, and outputs an abnormality sign alarm.

When the control device 50 outputs an abnormality sign alarm, it waits for the operator to replace the tool T (S220) or to give an instruction to rebuild the determination model 55 (instruction to continue using the tool T) (S220). S230). The operator who notices the abnormality sign alarm confirms the state of the tool T, replaces the tool T if necessary, and then inputs that the replacement of the tool T is completed via the operation panel 18. Alternatively, when the operator confirms the state of the tool T and determines that the replacement is not necessary yet, the operator gives an instruction to reconstruct the determination model 55 and continues to use the current tool T via the instruction screen of the operation panel 18. Enter the instructions and. When the control device 50 determines that the tool T has been replaced in S220, the control device 50 has been replaced with the same type of tool T, and it is not necessary to newly construct the determination model 55. Therefore, the control device 50 returns to S130 and performs processing.

Further, when the control device 50 determines that the reconstruction of the determination model 55 is instructed in S230, the control device 50 performs a process of reconstructing the determination model 55 including the vibration data when it is determined in S160 and S170 that there is a sign of abnormality. Then (S240), the process returns to S130. Here, FIG. 8 is an explanatory diagram showing a change in the similarity Ds with the model reconstruction. FIG. 8A shows a case where the similarity Ds accidentally falls below the value 0 at time T2, and the tool T is in a state where it can be continuously used. In this case, when the operator instructs the restructuring of the determination model 55, the control device 50 reconstructs the determination model 55 including the vibration data at the time T2. Then, since the control device 50 detects the sign of abnormality using the reconstructed determination model 55, it is possible to make it difficult for the similarity Ds to fall below the value 0 as shown in FIG. 8 (b). In this way, by reconstructing the determination model 55 by using the vibration data when the sign of abnormality is detected as normal data, it is possible to suppress the accidental drop in the similarity Ds and improve the detection accuracy of the sign of abnormality. Can be done. Further, even if a sign of abnormality is notified, the operator determines the continuous use of the tool T after confirming it, so that the replacement timing of the tool T can be made appropriate.

Here, the correspondence between the components of the present embodiment and the components of the present disclosure will be clarified. The control device 50 of the machine tool 10 of the present embodiment corresponds to the abnormality sign detection device, and the data collection unit 51 that collects the vibration data from the vibration sensor 36 in S110 of the abnormality sign detection process corresponds to the data collection unit. The model building unit 54 that builds the determination model 55 in S110 of the sign detection process corresponds to the model building unit, and the similarity calculation unit 56 that performs sign detection in S160 and S170 of the abnormality sign detection process corresponds to the sign detection unit. Further, the abnormality sign notification unit 58 that outputs the abnormality sign alarm by using the alarm 19 in S210 of the abnormality sign detection process corresponds to the sign notification unit. The threshold value changing unit 57 that accepts the threshold value Th setting change via the operation panel 18 in S130 and S140 of the abnormality sign detection process and changes the threshold value Th corresponds to the threshold value changing unit. The control device 50 that receives a reconstruction instruction via the operation panel 18 in S230 of the abnormality sign detection process corresponds to the reception unit. Further, the machine tool 10 corresponds to a machine tool.

In the machine tool 10 of the present embodiment described above, by constructing the determination model 55 in the 1-class SVM in the learning period (predetermined period) at the beginning of machining, the detection of a sign of abnormality of the tool T is promptly started. can do. Further, for another tool T used for the same machining, it is possible to detect a sign of abnormality from the beginning of machining by using the determination model 55 already constructed. Therefore, it is possible to detect a sign of abnormality of the tool T at an early stage and enable appropriate use of the tool T.

Further, the machine tool 10 can output an abnormality sign alarm of the tool T to urge the operator to take an appropriate response. Further, since the machine tool 10 can change the threshold value Th of the determination model 55, it is possible to adjust the detection sensitivity of the sign of abnormality. Further, since the machine tool 10 reconstructs the determination model 55 by including the vibration data when the sign of the abnormality of the tool T is detected in the data of the normal region based on the reconstruction instruction, the vibration data is accidental. It is possible to improve the accuracy of detecting a sign by suppressing the detection of a sign of abnormality when the area deviates from the normal region.

It goes without saying that the present disclosure is not limited to the above-described embodiment, and can be implemented in various embodiments as long as it belongs to the technical scope of the present disclosure.

For example, in the above-described embodiment, when the continuous use of the tool T is instructed when the sign of abnormality is output, the vibration data when the sign of abnormality is detected is included in the data of the normal region and the determination model 55 is re-executed. It was supposed to be built, but it is not limited to this. That is, the tool T may be continuously used and the determination model 55 may not be reconstructed. Alternatively, the tool T may be replaced, but the determination model 55 may be reconstructed.

In the above-described embodiment, the threshold Th setting change can be accepted from the operator, but the present invention is not limited to this, and the threshold Th setting change may not be accepted.

In the above-described embodiment, the control device 50 of the machine tool 10 detects a sign of abnormality, but the present invention is not limited to this, and a device other than the control device 50 of the machine tool 10 detects a sign of abnormality. And so on. For example, a management device that controls and manages a plurality of machine tools 10 may collect vibration data from each machine tool 10 and detect a sign of each abnormality. Further, in such a case, when a sign of abnormality is detected, the alarm is not limited to the one that outputs the warning of abnormality by using the alarm 19 provided in the machine tool 10, and the sign of abnormality is displayed on the display screen provided in the management device. A warning message to that effect may be displayed, or information indicating that a sign of abnormality has been detected may be transmitted from the management device to the worker's mobile terminal.

The abnormality sign detection device of the present disclosure may include a sign notification unit that notifies the effect of the abnormality sign (abnormality sign warning) when the sign of the tool abnormality is detected by the sign detection unit. In this way, it is possible to promptly notify the operator of the sign of an abnormality in the tool and prompt an appropriate response such as replacement of the tool.

The abnormality sign detection device of the present disclosure includes a threshold value changing unit that can change the threshold value of the normal region for determining whether or not the vibration data is included in the normal region based on an instruction from an operator. The sign detection unit may determine whether or not the vibration data is included in the normal region based on the changed threshold value changed by the threshold value changing unit. In this way, since the detection sensitivity of the sign of abnormality can be adjusted by changing the threshold value, it is possible to aim for earlier detection of the sign or suppress the detection of frequent sign.

The abnormality sign detection device of the present disclosure includes a reception unit that receives an instruction to reconstruct the determination model from an operator when a sign of abnormality of the tool is detected by the sign detection unit, and the model construction unit is provided. When the restructuring instruction is received by the reception unit, the determination model may be reconstructed by including the vibration data when a sign of abnormality of the tool is detected in the data of the normal region. By doing so, it is possible to suppress the detection of a sign of abnormality when the vibration data accidentally deviates from the normal region, so that the accuracy of detecting the sign can be improved.

The machine tool of the present disclosure is a machine tool that performs machining using replaceable tools, and is gisted to include any of the above-mentioned abnormality sign detection devices.

Since the machine tool of the present disclosure is provided with any of the above-mentioned abnormality sign detection devices, the same effect as that of the above-mentioned abnormality sign detection device, for example, the proper use of the tool can be made so that the sign of the tool abnormality can be detected at an early stage. The effect that is possible will be obtained.

The present disclosure can be used for monitoring abnormalities of tools used in machine tools.

10 Machine tools, 12 bases, 14 tables, 16 columns, 18 operation panels, 19 alarms, 20 work chuck devices, 22 mounts, 24 Y-axis moving mechanisms, 26 chuck claws, 30 spindle devices, 32 spindles, 34 spindle motors , 36 Vibration sensor, 40 Spindle movement mechanism, 42 Z-axis movement mechanism, 44 X-axis movement mechanism, 50 control device, 51 data collection unit, 52 vibration data area, 53 feature amount extraction unit, 54 model construction unit, 55 judgment model , 56 similarity calculation unit, 57 threshold change unit, 58 abnormality sign notification unit, Ds similarity, T tool, Th threshold, W work.

Claims (5)

An abnormality sign detection device that detects signs of abnormality in tools used for machining.
A data collection unit that collects vibration data of the tool during machining,
A model construction unit that constructs a judgment model in a one-class support vector machine using the vibration data collected by the data collection unit as data in a normal region during a predetermined period at the beginning of machining.
Using the determination model constructed by the model construction unit, a sign detection unit that detects the presence or absence of a sign of abnormality of the tool by determining whether or not the vibration data is included in the normal region.
An abnormality sign detection device equipped with. The abnormality sign detection device according to claim 1.
An abnormality sign detection device including a sign notification unit that notifies the effect of a sign of abnormality when a sign of an abnormality of the tool is detected by the sign detection unit. The abnormality sign detection device according to claim 1 or 2.
A threshold value changing unit capable of changing the threshold value of the normal region for determining whether or not the vibration data is included in the normal region is provided based on an instruction from an operator.
The sign detection unit is an abnormality sign detection device that determines whether or not the vibration data is included in the normal region based on the changed threshold value changed by the threshold value change unit. The abnormality sign detection device according to any one of claims 1 to 3.
When the sign detection unit detects a sign of abnormality of the tool, the reception unit is provided to receive an instruction to reconstruct the determination model from the operator.
When the reception unit receives the reconstruction instruction, the model construction unit reconstructs the determination model by including the vibration data when a sign of abnormality of the tool is detected in the data of the normal region. Abnormality sign detection device. A machine tool that performs machining using replaceable tools.
A machine tool comprising the abnormality sign detection device according to any one of claims 1 to 4.