JP7119337B2 - Tool life predictor - Google Patents

Description

The present invention relates to a tool life predicting device.

In Patent Document 1, a wear evaluation index is set in advance for each tool according to the degree of progress of wear of the tool that progresses due to the execution of each machining process, and a cumulative wear evaluation index that is accumulated according to the repetition of each machining process. is calculated for each used tool, and when the accumulated wear evaluation index reaches the limit index, it is determined that the used tool has reached the end of its life.

In Patent Document 2, the specific cutting resistance measured in advance for the workpiece to be machined this time, the already measured machining load value and the corresponding specific cutting resistance for the previously machined workpiece. is used to predict the machining load value expected to occur when machining the workpiece to be machined this time, and the predicted machining load value is used to detect whether the tool has reached the end of its life. .

In Patent Document 3, a correction coefficient is obtained that takes into account a material coefficient determined by the material to be processed and a processing condition coefficient based on the processing conditions, and the tool usage time is multiplied by the correction coefficient to correct the tool usage time. Cumulative hours of use are described to estimate tool life.

JP-A-2001-246534 Japanese Patent No. 5411055 JP-A-7-51998

Patent Document 1 can be applied when the same processing steps are repeated, but cannot be applied when different processing steps are performed. That is, the technique described in Patent Document 1 is applied to the mass production of workpieces of the same type. However, this technology cannot be applied to tool life prediction in high-mix low-volume production, that is, machining of high-mix low-volume workpieces.

Moreover, the state of the tool, such as the sharpness of the tool, changes from the start of use of the tool to the end of its service life. However, the techniques described in Patent Literatures 2 and 3 do not consider changes in the state of the tool. Therefore, with this technology, it is not possible to predict the tool life with high accuracy.

The present invention is a tool life prediction device that can predict tool life in high-mix low-volume production, and can predict tool life with high accuracy by considering the state of the tool from the start of use until the end of its life. intended to provide

The tool life prediction device according to the present invention is
A tool life prediction device that predicts the life of each of the plurality of types of tools when machining a plurality of types of workpieces with a plurality of types of tools,
When the plurality of types of workpieces W are subjected to the first machining using the first tools for information acquisition corresponding to each of the plurality of types of tools, grouping by type of the first tools, Further, the first machining information is acquired from the start of use of each of the first tools to the end of the life, and the first machining information is acquired when the first machining is performed for a plurality of the first tools. a first processing information acquisition unit to
a model storage unit that stores the computation model, which is a computation model determined based on the first machining information, for estimating the life of a tool of the same type as the first tool;
Second machining of machining the plurality of types of workpieces from the start of use of one of the second tools to the end of the life using a second tool that is the same type of tool as the first tool and whose life is to be predicted. a second processing information acquisition unit that acquires second processing information of the same type as the first processing information when performing the second processing, when performing the second processing;
a prediction unit that predicts the life of the second tool based on the second machining information and the arithmetic model;
with
The first machining information includes the usage time of the first tool, the characteristic amount of the torque of the motor of the spindle device that rotates the first tool, and the torque of the motor for relative movement between the first tool and the workpiece. feature quantity, feature quantity relating to vibration of the spindle device, machining conditions, and information relating to the material of the workpiece .

According to the present invention, the computation model is determined based on the first machining information from the start of use of the first tool to the end of its life when the first tool performs the first machining of a plurality of types of workpieces. there is In other words, the computational model considers the state of the first tool from the start of use until the end of the life of the first tool. Furthermore, the first machining information for determining the computational model is information when a plurality of types of workpieces are machined by the first tool. Therefore, the computational model is intended not for machining a specific type of workpiece, but for machining a plurality of types of workpieces. In other words, the calculation model is intended for high-mix low-volume production.

Then, the prediction unit uses the calculation model and second machining information when the second tool of the same type as the first tool performs the second machining, thereby predicting the life of the second tool. there is That is, by using the second machining information by the second tool and the calculation model, the prediction unit can obtain the state in which the second tool is positioned from the start of use to the end of its life. Life expectancy can be predicted.

As described above, according to the present invention, it is possible to predict the tool life in high-mix low-volume production, and it is possible to predict the tool life with high accuracy by considering the state of the tool from the start of use to the end of the tool life. Become.

It is a figure which shows the machining center which is production equipment. It is a figure which shows the functional block regarding production equipment. It is a functional block diagram of a tool life prediction device. It is a figure which shows reference|standard data. FIG. 4 is a diagram showing target data; FIG. 11 shows total combined data; FIG. 10 is a diagram showing tool-specific coupling data of group Gr1; FIG. 10 is a diagram showing tool-specific coupling data of group Gr2; FIG. 10 is a diagram showing tool-specific coupling data of group Gr3. FIG. 5 is a diagram showing the relationship between the number of machined workpieces and the tool usage time when a plurality of types of workpieces are machined. A diagram similar to FIG. 8 shows the acquisition timing of the first processing information. FIG. 9 is a diagram similar to FIG. 8 , showing levels corresponding to the remaining usage time predicted by the arithmetic model for each acquisition timing of the first processing information; FIG. 10 is a diagram showing a three-dimensional space in the case of using a machining frequency peak, a spindle torque peak, and an Overall value as variables as an example of variables of an arithmetic model; FIG. 3 is a diagram showing a two-dimensional space for explaining the k-nearest neighbor method as an example of a computational model; FIG. 4 is a diagram showing the level behavior of each tool according to the number of workpieces W machined. It is a figure for demonstrating the prediction of the timing of lifetime reach|attainment by a 2nd prediction part.

(1. Configuration of machine tool 1)
The machine tool 1 is a machine that sequentially processes a plurality of types of workpieces W based on a production plan. In other words, the machine tool 1 is a machine for high-mix low-volume production. The machine tool 1 uses tools to perform, for example, cutting, grinding, forging, electrical discharge machining, and the like. An example of the machine tool 1 will be described with reference to FIGS. 1 and 2. FIG. As an example of the machine tool 1, a horizontal machining center for cutting a workpiece W in a production line is taken as an example. As the configuration of the machining center as the machine tool 1, various known configurations can be adopted in addition to the configuration described below.

A machining center is configured, for example, as follows. A bed 11 is fixed to the installation surface, and a column 12 is supported by the bed 11 so as to be movable in the X-axis direction (front-rear direction on the paper surface of FIG. 1). A saddle 13 is supported on the front surface of the column 12 (right surface in FIG. 1) so as to be movable in the Y-axis direction (vertical direction in FIG. 1). A spindle device 14 is provided on the saddle 13 , and a tool 15 is held by a spindle (not shown) which is a rotating body in the spindle device 14 .

A table 16 is supported on the bed 11 so as to be movable in the Z-axis direction at a position facing the column 12 in the Z-axis direction (horizontal direction in FIG. 1). A workpiece W is fixed on the table 16 . Furthermore, a tool magazine 17 is provided on the side of the column 12 on the bed 11 . A plurality of tools 15 are stored in the tool magazine 17 . Furthermore, an automatic tool changer 18 is provided to replace a tool 15 with a designated tool number among the plurality of tools 15 stored in the tool magazine 17 with a tool 15 held in the spindle device 14. do.

Furthermore, as shown in FIG. 2, the machining center which is the machine tool 1 includes a CNC (Computerized Numerical Control) device 21, a PLC (Programmable Logic Controller) 22, various detectors 23a, 23b, 23c, 23d, 24a, 24b, 24c. , and a first prediction device 25 .

The CNC device 21 controls a motor (not shown) for rotating the spindle in the spindle device 14 according to a machining program (NC program), and controls the tool 15 attached to the spindle device 14 and the workpiece W. A motor (not shown) for relative movement is controlled. The CNC device 21 includes, for example, a detector 23a for detecting the driving current of the motor of the spindle device 14, a detector 23b for detecting the driving current of the motor for moving the column 12, the saddle 13 and the table 16, the X Information detected by the detector 23c for detecting the axial position, the Y-axis position of the saddle 13, the Z-axis position of the table 16, and the like is acquired, and each motor is controlled. The CNC device 21 also acquires information detected by the detector 23d that detects the outside air temperature, and performs, for example, thermal displacement correction.

The PLC 22 performs sequence control according to a ladder circuit, sequential function chart (SFC), or the like. The PLC 22 acquires information from ON/OFF detectors 24 a , 24 b , 24 c provided in the machine tool 1 . For example, the PLC 22 controls the operation for moving the tool 15 with the designated tool number to the replacement position in the automatic tool changer 18 and controls the operation of the automatic tool changer 18 . The PLC 22 also controls the supply of coolant by controlling the pump of a coolant device (not shown).

The first prediction device 25 (tool life prediction device) predicts the life of each of the multiple types of tools 15 when processing multiple types of workpieces W using multiple types of tools 15 . The first prediction device 25 detects data detected by detectors 23a-23d and 24a-24c provided in the machine tool 1 (corresponding to data relating to the machine tool 1), control data in the CNC device 21, I/ Acquire I/O data stored in the O memory.

Here, the first prediction device 25 is, for example, placed on the far side of the bed 11 or housed in a housing of a control panel (not shown). The first prediction device 25 is connected with the CNC device 21 and the PLC 22 . For example, the first prediction device 25 has a LAN connector, a USB connector, etc., and is connected to the CNC device 21 and the PLC 22 . That is, the first prediction device 25 is connected to the CNC device 21 and the PLC 22 via Ethernet (registered trademark), EtherCAT (registered trademark), or the like. That is, the first prediction device 25 communicates with the CNC device 21 and the PLC 22 using, for example, network protocols of the lower two layers (physical layer and data link layer) of the OSI reference model.

In general, the Internet protocol suite uses protocols of layer 3 (network layer) and above of the OSI reference model. The data transmission speed of the physical layer and data link layer is higher than the data transmission speed of the Internet protocol suite.

The first prediction device 25 in this embodiment constitutes edge computing connected by a lower-layer protocol close to communication targets (CNC device 21 and PLC 22). Edge computing is used as a name in contrast to cloud computing that uses the Internet protocol suite.

Also, the first prediction device 25 can be detachably connected to an input device 26 and a display device 27 as external devices. Therefore, the first prediction device 25 has a terminal for connecting to the display device 27 . The input device 26 inputs and edits the settings of the first prediction device 25 . The display device 27 can display the details of processing by the first prediction device 25 . Note that the first prediction device 25 may be configured to include the input device 26 and the display device 27 .

Furthermore, the first prediction device 25 connects the LAN cable connected to the LAN connector to a management device 28 (tool life prediction device) such as a server, so that the data such as the processing result by the first prediction device 25 is sent to the second It can be transmitted to the management device 28 as a prediction device. The management device 28 as a second prediction device stores the production plan and predicts the tool life using the processing result of the first prediction device 25, the production plan and the machining program. In other words, the management device 28 as the second prediction device predicts the life of each of the plurality of types of tools 15 in the case of machining the plurality of types of workpieces W using the plurality of types of tools 15 in consideration of the production plan. do.

Although the first prediction device 25 is described as a separate device from the CNC device 21 and the PLC 22, it can also be an embedded system such as the CNC device 21 and the PLC 22, and is arranged at a position separate from the machine tool 1. It can also be a personal computer, a server, or the like. Also, the first prediction device 25 and the management device 28 constitute a tool life prediction device 50 .

(2. Configuration of tool life prediction device 50)
The configuration of the tool life prediction device 50 will be described with reference to FIGS. 3 to 14. FIG. As shown in FIG. 3, the tool life prediction device 50 includes a first prediction device 25 and a management device 28 as a second prediction device.

The first prediction device 25 includes a first database 31, a reference data acquisition unit 32, a target data acquisition unit 33, a data acquisition determination unit 34, a combined data generation unit 35, a first processed information acquisition unit 36, a model storage unit 37, and , and a second processing information acquisition unit 38 .

The first database 31 acquires and stores preset types of data among data stored in the CNC device 21 and the PLC 22 . In the first database 31, data detected by the detectors 23a-23d and 24a-24c via the CNC device 21 and PLC 22 are stored. Further, the first database 31 stores data used for control by the CNC device 21 and I/O data used for control of the PLC 22 .

For example, in the case of continuously machining a plurality of workpieces with a machining center, data is stored in the first database 31 each time machining of one workpiece is completed. In other words, when one workpiece is finished to be machined, data on the workpiece is newly added to the first database 31 . Further, the data may be stored in a storage medium as a data file or the like, and the data may be added in one cycle such as one machining cycle, or acquired and added at regular intervals.

Here, it is assumed that the first database 31 includes reference data relating to the tool numbers (corresponding to "tool identification information") of the tools 15 currently in use (corresponding to "grouping criteria"). The reference data (File_A) relating to the tool 15 includes, as shown in FIG. 4, a tool number as identification information of the tool 15 and operation time information (including date and time) of the tool 15 . In this embodiment, as shown in FIG. 4, an example of the reference data includes time information when the tool 15 operates (time when the tool 15 starts machining) and identification information of the tool 15 being used at that time. including the tool number as The tool number of the tool 15 is acquired from the execution information of the machining program of the CNC device 21, the information regarding the tool change operation of the PLC 22, or the like.

A group in the reference data regarding the tool 15 is a group corresponding to each of the multiple types of tools 15 . That is, different types of tools 15 have different groups. In FIG. 4, new reference data is recorded at the timing when the tool 15 is changed. The timing at which the reference data is stored may be a predetermined cycle.

Further, the first database 31 includes target data regarding the state of the machine tool 1 detected by the detectors 23 a - 23 d and 24 a - 24 c provided on the machine tool 1 . For example, the first database 31 stores the usage time of the tool 15, the feature quantity of the torque of the motor of the spindle device 14, the feature quantity of the torque of the motors of the X, Y, and Z axes, the feature quantity related to the vibration of the spindle device 14, , machining conditions, the material of the workpiece W, and the like.

Here, the usage time of the tool 15 is the cumulative usage time of the tool 15 from the beginning of use. The characteristic amount of the torque of the motor of the spindle device 14 is, for example, a torque peak value, a variance value, an average value, or the like. The characteristic quantity of the torque of the motor on each of the X, Y, and Z axes is the torque peak value, variance value, average value, and the like. The feature values related to the vibration of the spindle device 14 are the machining frequency peak (corresponding to “spindle rotation speed × tool blade number”), rotation frequency peak (corresponding to “spindle rotation speed”), overall value (“FFT value of vibration value ), RMS value (corresponding to "variation in FFT value of vibration value"), and the like. Machining conditions include the number of revolutions of the spindle, the feed rate of each of the X, Y, and Z axes, the depth of cut, and the like.

For example, the first database 31 includes target data detected by the detector 23 a that detects the drive current of the motor of the spindle device 14 . As shown in FIG. 5, the target data (File_B) includes time information (including date and time) detected by the detector 23a and the drive current of the motor of the spindle device 14 obtained by the detector 23a at the time. including data and In this embodiment, an example of the target data includes, as shown in FIG. 5, the detected time (date and time) and the data of the drive current of the motor of the spindle device 14 acquired by the detector 23a at the detected time. . Target data is not information that can be grouped by itself. Also, the target data is stored at predetermined intervals. The sampling period for obtaining the tool number of the tool 15 is longer than the sampling period for the detection data of the drive current of the motor of the spindle device 14 .

The reference data acquisition unit 32 acquires reference data as a reference for grouping from data stored in the first database 31 . The target data acquisition unit 33 acquires target data from data stored in the first database 31 .

The data acquisition determination unit 34 determines that the reference data acquisition unit 32 has acquired the reference data and that the target data acquisition unit 33 has acquired the target data based on a preset algorithm. Here, when one workpiece has been machined, the first database 31 stores the reference data (File_A) and the target data (File_B). Therefore, the reference data acquisition unit 32 acquires the reference data (File_A) and the target data acquisition unit 33 acquires the target data (File_B) at the timing when one workpiece is finished being machined. In other words, the data acquisition determining unit 34 also determines that both data have been acquired at the timing when one workpiece is finished being machined.

The combined data generation unit 35 starts generating tool-specific combined data when the data acquisition determination unit 34 determines that the reference data (File_A) and the target data (File_B) have been acquired. The combined data by tool is the data detected in the target data (File_B) in the same time zone as the operation time zone of the reference data (File_A) for each type of tool 15 in the reference data (File_A) as the reference data (File_A). This is the combined data.

The combined data generation unit 35 includes a total combined data generation unit 35a and a tool-specific dividing unit 35b. The total combined data generation unit 35 a acquires the reference data (File_A) and the target data (File_B) from the reference data acquisition unit 32 and the target data acquisition unit 33 . Subsequently, as shown in FIG. 6, the total combined data generation unit 35a converts the acquired reference data (File_A) and target data (File_B) into operation time of the reference data (File_A) and detection time of the target data (File_B). Generate total combined data (File_C) that is associated and combined with .

That is, the total combined data (File_C) includes the time (date and time), the tool number of the tool 15 operating at that time, the drive current data detected at that time, and the like. At this time, as shown in FIG. 6, the tool number information at the time when the reference data (File_A) does not exist takes over the immediately preceding information. That is, the tool number remains T3 until the tool number is changed from T3 to T1. Also, the tool number remains T1 until the tool number is changed from T1 to T7.

The tool-by-tool dividing unit 35b divides the total combined data (File_C) by tool based on the types (Gr1, Gr2, Gr3) of the tools 15 of the reference data (File_A) in the total combined data (File_C). - Generate tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) for each type of tool 15, as shown in FIG. 7C. That is, each tool-specific combined data includes time (date and time), the tool number of the tool 15 that was operating at that time, and data on the drive current detected at that time. Then, the same tool-specific combined data has the same tool number.

The first machining information acquisition unit 36 performs the first machining on the plurality of types of workpieces W using the first tools for information acquisition corresponding to the plurality of types of tools 15, respectively. Acquire the first processing information from the start of use to the end of the service life. Here, the first processing information is tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) generated by the tool-specific division unit 35b when the first processing is performed.

The first machining is, for example, machining of a plurality of types of workpieces W that is executed based on an actual production plan for high-mix low-volume production. However, the first processing is not limited to processing based on an actual production plan, and can also be targeted for trial processing.

Here, when machining a plurality of types of workpieces W with a plurality of types of tools 15, each tool 15 is replaced with a new one each time it reaches the end of its life. For example, focusing on the tool 15 with the tool number T1, FIG. 8 shows the relationship between the number of processed workpieces W and the usage time of the tool 15 when the first machining is performed. In FIG. 8, when the tool 15 is replaced with a new one, the usage time of the tool 15 is reset. In other words, in FIG. 8, the ◯ mark means the timing at which the operator determines that the tool 15 has reached the end of its life and replaces the tool 15 with a new one.

As shown in FIG. 8, when the types of workpieces W are different, the number of workpieces W to be machined from the start of use of the tool 15 to the end of its life and the usage time of the tool 15 are different. For example, the usage time of the tool 15 that has reached the end of its life varies from 200 minutes or more to about 50 minutes. The number of workpieces W machined until the end of life is reached also varies.

Then, the first processing information acquisition unit 36 acquires the first processing information when the first processing is performed. The timing at which the first processed information acquisition unit 36 acquires the first processed information is indicated by, for example, the circle mark in FIG. FIG. 9, like FIG. 8, is a diagram showing the relationship between the number of machined workpieces W and the usage time of the tool 15. In FIG. The first machining information to be acquired is the tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) generated by the tool-specific dividing unit 35b. That is, the first machining information is information grouped according to the type of the tool 15, and includes the usage time of the tool 15, the characteristic amount of the torque of the motor of the spindle device 14, the motors of the X, Y, and Z axes. , the feature amount of the torque, the feature amount of the vibration of the spindle device 14, the machining conditions, the material of the workpiece W, and the like. Thus, the first processing information includes multiple types of variables.

The model storage unit 37 stores a computation model determined based on the first machining information about the plurality of types of workpieces W, and for estimating the life of a tool of the same type as the first tool. The calculation model is a model capable of predicting whether or not the tool 15 has reached the end of its life and the remaining usage time until the tool 15 reaches the end of its life.

For example, the calculation model may predict the remaining usage time itself until the end of the service life, or may determine a corresponding level from among a plurality of levels corresponding to the predicted remaining usage time. In this embodiment, the calculation model determines the latter level. As shown in FIG. 10, the multiple types of levels are classified into levels 1-5, for example. Level 1 corresponds to less than 15 minutes of expected usage time remaining. Level 2 corresponds to a predicted remaining usage time of 15 minutes or more and less than 30 minutes. Level 3 corresponds to a predicted remaining usage time of 30 minutes or more and less than 60 minutes. Level 4 corresponds to a predicted remaining usage time of 60 minutes or more and less than 120 minutes. Level 5 corresponds to an expected remaining usage time of 120 minutes or more. That is, level 1 is near the end of life, and level 5 is near new.

Then, the calculation model predicts a level corresponding to the remaining usage time based on the first machining information, and the actual remaining usage time of the tool 15 matches the remaining usage time predicted based on the first machining information. is set to FIG. 10 shows the level corresponding to the remaining usage time predicted using the set calculation model and the first processing information. As shown in FIG. 10, it can be seen that the level is 5 when the tool 15 is just replaced with a new one, and the level is 1 when the tool 15 is about to reach the end of its life.

Computational models include linear adaptation (e.g., linear adaptive control, etc.), nonlinear identification (e.g., sequential identification, etc.), Bayesian methods (e.g., naive Bayes classifiers, Bayesian networks, etc.), machine learning (e.g., neural networks, support vector machine (SVM), random forest, k-nearest neighbor method, etc.), regression analysis (eg, multiple regression analysis, ridge regression, logistic regression, etc.), etc. can be applied.

For example, as shown in FIG. 11, the computational model uses the machining frequency peak, the spindle torque peak and the Overall value as variables, and determines the level based on the values of these three variables. The computational model can be a model using four or more variables, or a model using one or two variables. Also, the model storage unit 37 may store a plurality of types of calculation models. For example, the model storage unit 37 may store a machine learning computation model and a regression analysis computation model. Also, the model storage unit 37 may store different calculation models according to, for example, the type of machining material and machining conditions.

Here, the calculation model uses the values of the variables at an arbitrary time to predict the life of the tool 15 at that arbitrary time. In other words, the computational model does not use past information to predict the life of the tool 15 at a specific time, but only uses information at that time (instantaneous). Note that the calculation model may use past information to predict the life of the tool 15 .

Further, the model storage unit 37 may update the computation model using the continuously acquired first processing information after storing the once determined computation model. Of course, the model storage unit 37 may not be updated after storing the calculation model once determined. When updating the calculation model, the first machining information acquisition unit 36 continues to acquire the tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) as the first machining information. On the other hand, if the calculation model is not updated, the first machining information acquisition unit 36 acquires the tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) as the first machining information after the calculation model is determined. no.

After the computational model is stored in the model storage unit 37, the second machining information acquisition unit 38 uses the second tool 15 of the same type as the first tool and whose life is to be predicted to obtain a plurality of types of workpieces. When the second processing for W is performed, the second processing information during the second processing is acquired. Here, the second processing information is tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) generated by the tool-specific division unit 35b when the second processing is performed.

The second machining is machining of a plurality of types of workpieces W that is executed subsequent to the first machining. In addition, when updating the computational model after the computational model is temporarily stored in the model storage unit 37, the first machining and the second machining are the same. That is, in the latter case, the tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) generated by the tool-specific division unit 35b are the first processing information and the second processing information. That is, the second machining information acquisition unit 38 acquires the tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3) as the second machining information, as indicated by the circles in FIG. Note that the second machining information, like the first machining information, is tool-specific combined data (File_D_Gr1, File_D_Gr2, File_D_Gr3), and therefore includes multiple types of variables.

The first prediction unit 39 predicts the life of the second tool based on the second machining information and the computational model. In particular, the first prediction unit 39 predicts the life of the second tool at an arbitrary time based on the second machining information and the calculation model only at an arbitrary time.

The first prediction unit 39 outputs information regarding life prediction according to the output format of the computation model. For example, if the computational model is a model for computing whether or not the life of the tool 15 has been reached, the first prediction unit 39 computes whether or not the life of the second tool has been reached. If the computation model is a model that computes the remaining usage time until the life of the tool 15 expires, the first prediction unit 39 computes the usage time until the second tool reaches the lifetime. When the computational model determines the level corresponding to the remaining usage time of the tool 15, the first predictor 39 determines the level of the second tool. In the present embodiment, the first prediction unit 39 uses an arithmetic model for determining levels 1 to 5 to determine the level of the second tool based on the second machining information.

For example, as shown in FIG. 11, the calculation model is a model that uses machining frequency peak, spindle torque peak, and overall value as variables, and determines the level based on the values of these three variables. In this case, as shown in FIG. 11, points corresponding to the first processing information are plotted in this three-dimensional space, and each plot point includes a level attribute.

Here, as an example of the calculation model, the k-nearest neighbor method is applied. In the k-nearest neighbor method, for example, in a three-dimensional space with three variables, the most frequently classified level is assigned. k is the set number. For clarity, reference is made to the two-dimensional space shown in FIG. As shown in FIG. 12, in a two-dimensional space, when a point corresponding to the second processing information at an arbitrary time is marked with a *, k (for example, 25) first Among the data groups corresponding to the processing information, there are the largest number of ● marks. Therefore, the point (* mark) corresponding to the second processing information at an arbitrary time is classified into ● mark, that is, level 4. That is, the second tool at an arbitrary time is determined to be level 4, that is, a level of 60 to 120 minutes of remaining usage time until reaching the end of its life.

Here, in the above description, the first prediction unit 39 determines the level of the second tool at each time using only one type of computation model. Here, when a plurality of types of calculation models are stored in the model storage unit 37, the first prediction unit 39 uses one type of calculation model selected from the plurality of types of calculation models to perform the second calculation. You may make it determine the level of a tool. At this time, the calculation model may be selected according to, for example, the material of the workpiece W and the machining conditions.

Further, when a plurality of types of calculation models are stored in the model storage unit 37, the first prediction unit 39 uses each of the plurality of types of calculation models to determine the level of each of the second tools. may In this case, since a plurality of levels are obtained, the highest level among the plurality of obtained levels can be used as the level of the second tool. In this case, all of the plurality of types of calculation models stored in the model storage unit 37 may be used, or a plurality of types of calculation models selected from among the plurality of types of calculation models stored may be used. can be

As described above, the first prediction unit 39 determines the level of the second tool at each time (each instant). The first prediction unit 39 can also determine whether or not the second tool has reached the end of its life based on the level behavior of the second tool. This will be described with reference to FIG.

FIG. 13 shows the level behavior according to the number of workpieces W machined for each tool 15 with tool numbers T1 to T50. When the number of machined workpieces W is 0, it is time to start using the tool. For example, according to the level behavior of the tool 15 with the tool number T1, the number of workpieces W machined from the start of use of the tool 15 with the tool number T1 (hereinafter referred to as "the number of machining") is 0 to about 50. Until then, the tool 15 with tool number T1 is level 5. After that, the tool 15 with the tool number T1 alternates between the level 5 and the level 4 when the number of machining is about 50 to about 60. After that, the tool 15 with the tool number T1 is level 4 from about 60 to about 80 pieces to be machined.

After that, the tool 15 with the tool number T1 alternates between level 4 and level 3 when the number of machining is about 80 to 90. After that, when the number of machining is about 95, the tool 15 with the tool number T1 changes from level 4 to level 2 at once. After that, the tool 15 with the tool number T1 alternates between the level 2 and the level 1 when the number of machining is about 95 to 120.

For example, the life of the tool 15 with the tool number T1 is reached when a predetermined period of time has passed since the tool 15 with the tool number T1 reached level 1. However, as shown in FIG. 12, even if it reaches level 1, it may return to level 2. Therefore, the first prediction unit 39 predicts the end of the life of the second tool based on the number of times or frequency that the determined level is the level (level 1) that minimizes the remaining usage time. For example, the first prediction unit 39 predicts that the second tool has reached the end of its life when the number of times of level 1 reaches ten. Further, the first prediction unit 39 may predict that the second tool has reached the end of its life when the number of times of level 1 reaches 5 out of a predetermined number of processes (for example, 10 times). . In addition to the number of times and frequency, the first prediction unit 39 may predict that the second tool has reached the end of its life when a predetermined number of processes has been reached after reaching level 1 for the first time. good.

The first prediction unit 39 sends information about the tool 15 that has reached the end of its life or the tool 15 that is about to reach the end of its life to the operation panel (not shown) of the machine tool 1, a mobile terminal owned by the worker, etc. Execute notification processing and display processing. The first prediction unit 39 can also execute processing for displaying the current level of each tool 15 with the tool numbers T1 to T50 on the operation panel, mobile terminal, or the like. In addition to the current level, the first prediction unit 39 can also execute processing for displaying the behavior of levels up to the current time, as shown in FIG. 12 . Therefore, the operator can grasp which tool 15 has reached the end of its life, whether or not there is a tool 15 that is about to reach the end of its life, and the behavior until the end of its life. Therefore, the operator can prepare in advance the tool 15 to be replaced.

The tool life prediction device 50 includes a management device 28 as a second prediction device in addition to the first prediction device 25 . The management device 28 is connected to the first prediction device 25 via a LAN cable or the like, as described above. The management device 28 is composed of a production plan storage section 41 , a machining program storage section 42 and a second prediction section 43 .

The production plan storage unit 41 stores production plans for multiple types of workpieces W to be machined by the machine tool 1 . The production plan includes information relating to the order of machining of multiple types of workpieces W, the time to start machining each workpiece W, and the like. The machining program storage unit 42 stores a machining program corresponding to the workpiece W to be machined. Although not shown, the machine tool 1 processes the workpiece W according to the production plan, and uses a machining program stored in the machining program storage unit 42 when machining the workpiece W. However, the production plan and machining program may be acquired in advance from the management device 28 and stored by the controller (not shown) of the machine tool 1 .

The second prediction unit 43 acquires the remaining usage time until the end of the life of each tool 15 as the second tool predicted by the first prediction unit 39 . Furthermore, the second prediction unit 43 predicts the timing when each tool 15 will reach the end of its life by using the obtained remaining usage time of each tool 15, the production plan, and the machining program. Prediction of the end of life timing by the second prediction unit 43 will be described with reference to FIG. 14 .

In FIG. 14, the horizontal axis represents the processing order of the workpiece W based on the production plan, and the plan for each tool 15 with the tool numbers T1 to T4 when each workpiece W is processed by the tools 15 with the tool numbers T1 to T4. The vertical axis represents usage time. The second prediction unit 43 grasps the processing order of the workpiece W from the production plan stored in the production plan storage unit 41 . Further, the second prediction unit 43 can predict the time during which each tool 15 is used in the machining program by analyzing the machining program stored in the machining program storage unit 42 . Specifically, the second prediction unit 43 predicts the time that each tool 15 will be used based on the distance of the machining path by each tool 15 and the moving speed of each tool 15 during machining. be able to. Hereinafter, the time during which each tool 15 is used will be referred to as planned use time.

Here, it is assumed that the first prediction unit 39 determines that the remaining usage time until the tool 15 with the tool number T3 reaches the end of its service life is 15 minutes. For the first and second workpieces W, the tools 15 with tool numbers T1-T4 are not used. For the third workpiece W, the tools 15 with tool numbers T1 and T2 are not used, and the tools 15 with tool numbers T3 and T4 are used. At this time, the planned usage times of the tools 15 with tool numbers T3 and T4 are 3 minutes and 1 minute, respectively. For the fourth workpiece W, the tools 15 with tool numbers T1-T4 are used, and the planned usage times of the tools 15 with tool numbers T1-T4 are 27 minutes, 20 minutes, 37 minutes, 4 minutes, respectively. minutes. For the fifth to seventh workpieces W, all of the tools 15 with tool numbers T1 to T4 are used.

Focusing on the tool 15 with the tool number T3, when the first to third workpieces W are machined, the total planned usage time of the tool 15 with the tool number T3 is 3 minutes. When the first to fourth workpieces W are machined, the total planned usage time of the tool 15 with the tool number T3 is 39 minutes. Since the remaining usage time of the tool 15 with the tool number T3 is 15 minutes, the second prediction unit 43 predicts that the tool 15 with the tool number T3 will reach the end of its life while the fourth workpiece W is being processed.

However, since the prediction by the second prediction unit 43 described above uses only the usage time of the tool 15, it does not consider the material of the workpiece W, the machining conditions, etc., and is not accurate. Therefore, the second prediction unit 43 may correct the planned usage time according to the material of the workpiece W, processing conditions, and the like.

The second prediction unit 43 obtains the timing of the end of the life of each tool 15, that is, the information about which workpiece W is to be processed when the end of the life is reached, from the operation panel of the machine tool 1, the portable terminal owned by the worker, or the like. Send to Since the operation panel and the portable terminal can display the information, the worker can prepare the tool 15 to be replaced in advance.

The first prediction section 39 and the second prediction section 43 correspond to the prediction section 60 in the present invention. Although the first prediction unit 39 and the second prediction unit 43 are separate devices in the above description, they may be integrated. That is, the second prediction unit 43 may be provided in the machine tool 1 . In addition to the second prediction unit 43 , the production plan storage unit 41 , the machining program storage unit 42 and the second prediction unit 43 described as the management device 28 may be provided in the machine tool 1 . As with the first prediction unit 39, the second prediction unit 43 can also be an embedded system such as the CNC device 21 or the PLC 22. You can also

(3. Effect)
The calculation model stored in the model storage unit 37 is based on the first machining information from the start of use of the first tool to the end of its life when the first tool performs the first machining of a plurality of types of workpieces W. has been decided. In other words, the computational model considers the state of the first tool from the start of use until the end of the life of the first tool. Furthermore, the first machining information for determining the computational model is information when a plurality of types of workpieces W are machined by the first tool. Therefore, the calculation model does not target the case where a specific type of workpiece W is machined, but targets the case where a plurality of types of workpieces W are machined. In other words, the calculation model is intended for high-mix low-volume production.

Then, the first prediction unit 39 uses the calculation model and the second machining information when the second tool of the same type as the first tool performs the second machining, thereby predicting the life of the second tool. It is carried out. That is, the first prediction unit 39 can obtain the position of the second tool from the start of use to the end of its life by using the second machining information of the second tool and the calculation model. It is possible to predict the life of two tools.

As described above, it is possible to predict the life of the tool 15 in high-mix low-volume production, and it is possible to predict the life of the tool 15 with high accuracy by considering the state of the tool 15 from the start of use to the end of the life.

In FIG. 10 , the first machining information used for determining the computational model targets 16 tools 15 , but may target two or more tools 15 . However, the greater the number of samples, the higher the accuracy of the computational model.

Further, the first prediction unit 39 may predict whether or not the service life will be reached, the remaining usage time until the service life is reached, or the level as the service life prediction. It is preferable to determine the target to be predicted by the first prediction unit 39 according to the purpose.

Also, the first prediction unit 39 predicts the life of the tool 15 at the arbitrary time using the second machining information only at the arbitrary time. That is, the first prediction unit 39 does not use past information when predicting the life. In other words, the calculation model is a model that can predict the service life using only the information at that moment. As a result, it is possible to simplify the calculation model and shorten the processing time required for the life prediction, so that the life can be predicted in real time.

Here, using multiple types of variables can realize more accurate life prediction. However, if life prediction is performed using a plurality of types of variables while using past information, the computation model becomes complicated and the processing time required for life prediction increases. This makes it impossible to predict the service life in real time. However, although the first prediction unit 39 uses a plurality of variables, the life prediction is performed using the second machining information only at an arbitrary time, so real-time and highly accurate life prediction is possible. Become.

The first prediction unit 39 may use one type of calculation model, or may use a plurality of types of calculation models. By preparing a plurality of kinds of calculation models, it is possible to select the calculation model to be applied according to the material of the workpiece W, for example. As a result, highly accurate life prediction is possible. It is also possible to comprehensively judge the final life expectancy based on a plurality of life expectancy predictions using a plurality of kinds of computation models. Even in this way, highly accurate life prediction is possible.

Also, the second prediction unit 43 predicted the timing when the life of the tool 15 is reached using the production plan and the machining program. This makes it possible to predict how many workpieces W can be machined using the current tool 15 . Therefore, it becomes possible to use the tool 15 until the end of its life.

Also, the first machining information and the second machining information are united data by tool (File_D_Gr1, File_D_Gr2, File_D_Gr3) for each type of tool 15 . Normally, it is not easy to collect various information for each type of tool 15 . However, the combined data generator 35 generates the tool-specific combined data using the reference data including the identification information of the tool 15 and the operation time information of the tool 15 as a criterion for grouping. Therefore, as the first machining information and the second machining information, it is possible to reliably obtain information in which many pieces of information are associated with each type of tool 15 . As a result, a wide variety of computational models can be easily used, and a computational model capable of highly accurate life prediction can be easily determined.

Moreover, RFID (RF tag) may be attached to the tool 15 and the workpiece W (including the transport box) to discriminate the type of the tool 15 and the workpiece W and to reflect it in the prediction.

1: machine tool, 15: tools (first tool, second tool), 23a, 23b, 23c, 23d, 24a, 24b, 24c: detector, 25: first prediction device, 28: management device, 32: reference Data acquisition unit, 33: target data acquisition unit, 35: combined data generation unit, 36: first processing information acquisition unit, 37: model storage unit, 38: second processing information acquisition unit, 39: first prediction unit (prediction part), 41: production plan storage unit, 42: machining program storage unit, 43: second prediction unit (prediction unit), 50: tool life prediction device, 60: prediction unit, W: workpiece

Claims (9)

A tool life prediction device that predicts the life of each of the plurality of types of tools when machining a plurality of types of workpieces with a plurality of types of tools,
When the plurality of types of workpieces W are subjected to the first machining using the first tools for information acquisition corresponding to each of the plurality of types of tools, grouping by type of the first tools, Further, the first machining information is acquired from the start of use of each of the first tools to the end of the life, and the first machining information is acquired when the first machining is performed for a plurality of the first tools. a first processing information acquisition unit to
a model storage unit that stores the computation model, which is a computation model determined based on the first machining information, for estimating the life of a tool of the same type as the first tool;
Second machining of machining the plurality of types of workpieces from the start of use of one of the second tools to the end of the life using a second tool that is the same type of tool as the first tool and whose life is to be predicted. a second processing information acquisition unit that acquires second processing information of the same type as the first processing information when performing the second processing, when performing the second processing;
a prediction unit that predicts the life of the second tool based on the second machining information and the arithmetic model;
with
The first machining information includes the usage time of the first tool, the characteristic amount of the torque of the motor of the spindle device that rotates the first tool, and the torque of the motor for relative movement between the first tool and the workpiece. feature quantity, feature quantity relating to vibration of the spindle device, machining conditions, and information relating to the material of the workpiece . The first processing is processing in high-mix low-volume production,
2. The tool life prediction device according to claim 1, wherein said arithmetic model is determined based on said first machining information when machining is performed in high-mix low-volume production. The arithmetic model is capable of predicting whether or not the service life has been reached and the remaining usage time until the service life is reached,
3. The tool life prediction according to claim 1 or 2 , wherein the prediction unit calculates whether or not the second tool has reached the end of its life and the remaining usage time until the end of the life of the second tool, based on the second machining information and the arithmetic model. Device. 4. The predicting unit according to any one of claims 1 to 3 , wherein the prediction unit predicts the life of the second tool at the arbitrary time based on the second machining information and the arithmetic model only at the arbitrary time. tool life predictor. The first processing information and the second processing information include multiple types of variables,
5. The tool life prediction device according to claim 4 , wherein the arithmetic model uses the plurality of types of variables to predict the life of the tool at the arbitrary time. The model storage unit stores a plurality of types of the computational models determined based on the first processing information,
6. The predicting unit according to any one of claims 1 to 5 , wherein the prediction unit predicts the life of the second tool based on one or a plurality of types of the calculation model selected from the plurality of types of the calculation model. The tool life prediction device according to the item. 7. The method according to any one of claims 1 to 6 , wherein the prediction unit predicts the timing when the second tool reaches the end of its life using the production plan for the plurality of types of workpieces and the machining program for the plurality of types of workpieces. The tool life prediction device according to any one of items. The arithmetic model determines a corresponding level from among a plurality of levels corresponding to the remaining usage time until the end of life,
The prediction unit determines the level of the second tool based on the second machining information and the arithmetic model, and the number of times the determined level is the level that minimizes the remaining usage time. The tool life prediction device according to any one of claims 1 to 7 , which predicts the end of life of said second tool based on the frequency. The first processing information acquisition unit and the second processing information acquisition unit are
In a machine tool, a reference data acquisition unit that acquires reference data that is reference data for grouping and that includes identification information of the tool and operation time information of the tool;
a target data acquisition unit that acquires target data relating to the state of the machine tool detected by a detector provided in the machine tool;
For each type of tool in the reference data, data detected in the target data in the same time zone as the operation time zone of the reference data is combined with the reference data to generate tool-specific combined data, and a combined data generation unit that uses separate combined data as the first processed information or the second processed information;
The tool life prediction device according to any one of claims 1 to 8 , comprising:

Images