JP7121623B2 - State estimation system, state estimation method, and service provision method - Google Patents

Description

Application of Article 30, Paragraph 2 of the Patent Law [Meeting name] 28th (Hitachi Industrial Equipment) seminar [Date] February 9, 2018 [Publisher name] Hitachi, Ltd. [Publication name] News Release [ Publication date: June 26, 2018 Meeting name: (Kenkai) IR Date: June 28, 2018

The present invention relates to a state estimation system, a state estimation method, and a service provision method.

2. Description of the Related Art Conventionally, processing devices such as electric presses that use dies to perform punching, bending, molding, shearing, and cutting are known.

With this type of machining apparatus, the mold deteriorates as machining is repeated, and the finished quality of the object to be machined begins to deteriorate. If the processing is repeated as it is, processing defects such as burrs will occur. Once a processing defect occurs, inspection work is performed after processing, and the defective product is disposed of, which is time-consuming and costly.
Therefore, it is desirable to appropriately manage signs of deterioration of the mold.

Patent Document 1 discloses, as a conventional technique for this purpose, "a bottom dead center detection sensor that detects the lowest point in the pressing direction of the mold 20 and an AE (Acoustic Emission) sensor that detects elastic waves generated from the processed portion of the mold 20. Install the sensor on the target device. From the output values of these sensors, a value obtained by averaging the lowest point and a value of a predetermined frequency component obtained by FFT processing the elastic wave are obtained, and an abnormality prediction score of the mold 20 is obtained from these values.” .

Japanese Patent Application Laid-Open No. 2018-62000

In the technique disclosed in Patent Document 1, a bottom dead center detection sensor and an AE sensor are indispensable, and the system configuration becomes complicated accordingly.

Accordingly, it is an object of the present invention to provide a technique for estimating the state of deterioration of a target device that does not require a sensor.

In order to solve the above problems, one representative state estimation system of the present invention is a state estimation system for estimating the deterioration state of a target device, wherein the control unit of the target device controls the drive unit of the target device. A data acquisition unit capable of acquiring the control internal value to be used, and estimating the time required for at least a part of the drive process of the target device based on the time-series change of the control internal value, and evaluating the period based on the required time A period estimating unit that obtains a value, an estimator that stores an estimation rule for the value of the period evaluation value and the deterioration of the target device, and a value of the period evaluation value obtained by the period estimating unit that is applied to the estimation rule of the estimator. and a degradation estimator for estimating the degradation of the target device.

The present invention makes it possible to estimate the deterioration state of a target device without requiring a sensor.
Problems, configurations, and effects other than those described above will be clarified by the following description of the embodiments.

It is a figure explaining a system configuration. It is explanatory drawing explaining each process in a punching process. FIG. 4 is a diagram showing an example of time-series data of control internal values; FIG. 4 is a diagram for explaining the influence of deterioration of the mold 20 on the object to be processed; FIG. 4 is an explanatory diagram for explaining changes in time-series data due to mold deterioration; It is a figure which shows the relationship between a deterioration degree and a required period. It is a figure which shows the correct condition of an estimated deterioration degree. 4 is a flow chart describing the operation of the state estimation system; It is a figure explaining the variation of a required period. It is a figure explaining management of the state estimation system by a cloud server.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a diagram for explaining the system configuration.
In the figure, a target device 100 such as an electric press machine is configured with a drive section 101 and a control section 102 .

The drive unit 101 includes a servomotor 1, a servo amplifier 2 that drives the servomotor 1, an eccentric shaft 11 that is rotationally driven by the servomotor 1, a transmission shaft 12 that transmits the motion of the eccentric shaft 11, and a slide 13 that converts the motion transmitted by the transmission shaft 12 into vertical motion.

The vertical movement of the slide 13 is transmitted to one upper mold 21 of the mold 20 . The other lower mold 22 of the mold 20 is fixed at a position for receiving the upper mold 21 . Objects 30 to be processed are sequentially inserted between the molds 20 (upper mold 21 and lower mold 22) by a transport mechanism (not shown). In synchronism with this insertion of the workpiece 30 , the slide 13 pushes the mold 20 to process the workpiece 30 .

The control unit 102 controls the finishing of the workpiece 30 after processing by controlling the driving unit 101 so as to comply with preset processing conditions.

On the other hand, the state estimation system 200 includes a data acquisition section 201 , a period estimation section 202 , an estimator 203 , a deterioration estimation section 204 , a mold state notification section 205 and a control change section 210 .

The data acquisition unit 201 acquires the control internal values 102a from the control unit 102 in time series. A case in which a control command current value (torque command value generated within the control unit 102 to control the motor torque of the servomotor 1) is used as the control internal value 102a will be described below.

The period estimator 202 analyzes time-series changes in the control internal value 102a in the time domain, and estimates the required period 202a required for at least part of the processing using the mold 20 as a period evaluation value.

The estimator 203 stores rules for estimating the length of the required period 202a and the degree of deterioration of the mold 20 (including signs of deterioration).

The deterioration estimator 204 applies the length of the required period 202a estimated by the period estimator 202 to the estimation rule of the estimator 203 to estimate the mold state.

The mold state notification unit 205 notifies (the terminal of) the person in charge of maintenance of the mold 20 of the deterioration of the mold 20 estimated by the deterioration estimation unit 204 .

The control change unit 210 acquires the mold state (estimated degree of deterioration) estimated by the state estimation system 200, and changes the control command (processing conditions, etc.) to the control unit 102 according to the estimation, so that the mold Realize processing that changes by adapting to the situation.

<Description of punching process>
Next, taking punching as an example, the relationship between the process and the time series change of the control internal value 102a will be described.
FIG. 2 is a diagram showing the punching process of the target device 100 divided into five steps A to E. FIG.
FIG. 3 is a diagram showing time-series changes in the control internal value 102a (here, the control command current value) in steps A to E. FIG. Steps A to E here correspond to driving steps of the target device 100 .

[Process A] This is the initial stage when the control unit 102 is activated. The slide 13 shown in FIG. 2A is at the origin position (top dead center) and is not driven, and the control internal value 102a at point A shown in FIG. 3 maintains the initial value.

[Step B] The controller 102 starts rotating the servomotor 1 . As shown in FIG. 2B, the slide 13 begins to descend. At this point, the pressing force of the slide 13 is not transmitted to the mold 20 (upper mold 21) and the workpiece 30, and the torque of the servomotor 1 is small. Therefore, as shown in FIG. 3, the control internal value 102a at point B falls within the steady-state value range.

[Steps B to C] As shown in FIGS. 2B to 2C, the slide 13 comes into contact with the mold 20 (upper mold 21) and starts pushing downward. However, the pressing force is not transmitted to the workpiece 30, and the torque of the servomotor 1 remains small. Therefore, the control internal value 102a from point B to point C shown in FIG. 3 maintains a range of steady values.

[Step C] As shown in FIG. 2C , the slide 13 and the mold 20 (upper mold 21 ) are united and moved downward to come into contact with the workpiece 30 . The position of the workpiece 30 is fixed by the mold 20 (lower mold 22). In this state, the torque of the servomotor 1 starts to rise in order to apply a pressing force to the workpiece 30 . Therefore, the control internal value 102a starts increasing from the steady value at point C shown in FIG.

[Steps C to D] In FIGS. 2C to 2D, the workpiece 30 is sandwiched between the molds 20 and pressed to cause deformation (elastic deformation→plastic deformation). During this time, the controller 102 continuously increases the torque of the servomotor 1 . Therefore, the control internal value 102a continues to rise from point C to point D shown in FIG.

[Step D] In FIG. 2[D], a crack 31 is generated in the workpiece 30 after continuing plastic deformation. The torque applied to the servomotor 1 reaches its maximum peak at the moment (immediately before) the crack 31 occurs. Therefore, the control internal value 102a reaches a peak value at point D shown in FIG.

[Steps D to E] In FIGS. 2D to 2E, the machining reaction force is released as the crack 31 progresses, so the torque applied to the servomotor 1 decreases. Therefore, the control internal value 102a decreases from point D to point E shown in FIG.

[Step E] FIG. 2E shows the moment when the workpiece 30 is completely broken and separated from the scrap 30a. At this time, the machining reaction force from the workpiece 30 completely disappears, so the torque applied to the servomotor 1 becomes the minimum value. Therefore, the control internal value 102a becomes the local minimum value at point E shown in FIG.

When the series of punching operations described above is completed, the control internal value 102a returns to the range of steady values. The slide 13 rises together with the upper mold 21, and returns to the origin position (top dead center) at point F shown in FIG. After that, the motor stops rotating, so that the control internal value 102a returns to the initial value.

<Influence of mold deterioration>
FIG. 4 is a diagram for explaining the influence of deterioration of the mold 20 (upper mold 21, lower mold 22) on the workpiece 30. As shown in FIG.
FIG. 4(a) is a schematic diagram when the upper die 21 and the lower die 22 are normal. In this case, the fracture surface 32 is generated at a position connecting the sharp corner of the upper mold 21 and the contact portion of the workpiece 30 and the sharp corner of the lower mold 22 and the contact portion of the workpiece 30 .

On the other hand, as shown in FIG. 4B, at the deteriorated and blunt ends 21a and 22a, the force is not concentrated on the workpiece 30, and the crack generation position is not stable. Furthermore, the time (point D shown in FIG. 3) until the crack occurs also tends to be delayed.

Furthermore, since the force that propagates the crack is not efficiently concentrated at the blunt end 21a and the end 22a, the fractured surface 32a does not proceed linearly, the fracture distance is long, the fracture speed is slow, and the crack is cracked. The time required from the occurrence to the completion of fracture is long. Further, if the crack development deviates between the upper mold 21 and the lower mold 22 beyond the limit, burrs (processing defects) are likely to occur on the machined end face of the machined object 30 .

<State estimation of mold 20>
FIG. 5 is an explanatory diagram for explaining differences in time-series data due to mold deterioration.

In the same figure, the first time points t1 (D1) and t1 (D2) are the peak values (corresponding to point D in FIG. 3) of the control internal value 102a when working with normal molds and deteriorated molds, respectively. This is the result. In the same figure, the second time points t2 (E1) and t2 (E2) are the results of measuring the end value of the completion of breaking (corresponding to point E in FIG. 3).
Since it is affected by the mold deterioration as described above, the required period (t2-t1) from the first time to the second time becomes longer as the mold deteriorates.

FIG. 6 is a diagram showing the relationship between the required period (t2-t1) and the measured value of the degree of deterioration (the amount of wear at the ends of the upper die 21 and the lower die 22).

In the figure, the required period and the degree of deterioration show a strong positive correlation.
By performing regression analysis (approximation by a regression formula) on this correlation, a mold state estimation formula can be obtained. This mold state estimation formula is stored in the estimator 203 as an estimation rule in the form of coefficient parameters and a data table.

By applying the required period from the peak point D to the end point E of the control internal value 102a to the mold state estimation formula stored in the estimator 203, the estimated degree of deterioration can be obtained.

FIG. 7 is a diagram comparing the estimated degree of deterioration and the actual degree of deterioration.

According to the figure, it can be seen that the estimated degree of deterioration is close to the correct answer. In particular, in areas where the degree of deterioration is low (areas showing signs of mold deterioration), the variation in correct answers does not increase. Therefore, the use of the required period is suitable for monitoring slight deterioration (symptoms) of the mold 20 .

<Flow of mold state estimation>
FIG. 8 is a flow chart describing the operation of state estimation system 200 .
The mold state estimation operation will be described along the step numbers in the figure.

Step S10: The data acquisition unit 201 waits for the activation of processing by the control unit 102 (to the NO side of step S10). When the activation of the processing process is detected, the data acquisition unit 201 shifts the operation to step S11.

Step S11: The data acquisition unit 201 retrieves data for one cycle of processing (from point B to point F in FIG. The control internal values 102a corresponding to the time period until 10:00 a.m. are acquired in chronological order. Note that the data acquisition unit 201 performs this operation multiple times, and performs filter processing such as averaging and median processing among the multiple pieces of time-series data thus obtained, thereby suppressing noise components such as torque ripple. may The data acquisition unit 201 outputs the obtained time-series data of the control internal value 102 a to the period estimation unit 202 .

Step S12: The period estimator 202 searches for the maximum peak value (point D in FIG. 3) from the time-series data of the control internal value 102a, and obtains the time of point D as the first time point t1(D). This first time point t1(D) is presumed to be the time point when the crack 31 occurs in the workpiece 30 .

Step S13: The period estimating unit 202 searches for a local minimum value (point E in FIG. 3) from the time-series data of the control internal value 102a after the first time point t1(D), and calculates the time of point E as the 2 Time point t2(E). This second time t2(E) is presumed to be the time when the workpiece 30 is completely broken.

Step S14: The period estimator 202 obtains the time difference (t2-t1) between the second time point t2 (E) and the first time point t1 (D), and defines it as the required period. This required period is presumed to be the required period from the occurrence of cracks, which is part of the processing, to the completion of breakage. Period estimating section 202 outputs the estimated required period to deterioration estimating section 204 as a period evaluation value.

The important point here is that a dedicated sensor is not provided to monitor the occurrence of cracks and the completion of breakage, and based on the time-series change in the control internal value 102a that can be obtained from the control unit 102, the crack occurrence and the completion of breakage are detected. It is a point (sensorless) that estimates the time required to reach the point.

Step S15: The deterioration estimator 204 applies the required period from the occurrence of cracks to the completion of breakage (equivalent to the period evaluation value) to the estimation rule (here, mold state estimation formula) of the estimator 203, thereby obtaining the estimated degree of deterioration. to estimate The estimation rule used here changes according to the type of mold 20 and processing conditions. Therefore, it is preferable to adaptively switch to an estimation rule according to the type of mold 20 and processing conditions.

Step S16: The deterioration estimation unit 204 stores the estimated deterioration degree as a history together with the current date and time. Further, the deterioration estimation unit 204 notifies (the terminal of) the person in charge of maintenance of the mold 20 of the estimated degree of deterioration via the mold state notification unit 205 . A person in charge of maintenance (or a terminal) can prepare a replacement mold 20 in advance by predicting the replacement time of the mold 20 based on the estimated degree of deterioration and history information.

Step S17: The mold state notification unit 205 determines whether the estimated degree of deterioration is large or small, and determines that there is a sign (possibility) of defective processing when the estimated degree of deterioration exceeds a predetermined threshold. Since the threshold used for this determination changes depending on the type of the mold 20 and the processing conditions, it is preferable to switch adaptively according to the type of the mold 20 and the processing conditions.

Also, in the history of the estimated degree of deterioration, if there is a sudden change in the estimated degree of deterioration that exceeds natural wear, it is determined that the mold 20 has an abnormality (such as chipping or adhesion of processing waste).

The mold state notification unit 205 determines that the mold 20 needs to be replaced or repaired in either case of a defect symptom or abnormality, and the operation proceeds to step S18.
Otherwise, the mold state notification unit 205 shifts the operation to step S19.

Step S18: The mold state notification unit 205 instructs the control change unit 210 to stop the target device 100. FIG. After that, the mold state notification unit 205 warns (the terminal of) the person in charge of maintenance of the mold 20 of the “symptom of defect or abnormality of the mold 20 ” together with the identification ID of the target device 100 . The person in charge of maintenance replaces or repairs the mold 20 for the target device 100 indicated by the identification ID according to this warning.
(The stopped target device 100 is restarted by a maintenance person who has finished replacing or repairing the mold 20. With this restart, the state estimation system 200 resumes operation from step S10.)

Step S<b>19 : The mold state notification section 205 outputs the estimated degree of deterioration to the control change section 210 . The control change unit 210 determines whether or not to change the control of the control unit 102 according to the value of the estimated degree of deterioration and the change in the history, and if so, determines whether processing is prioritized or die is prioritized.

For example, if the estimated degree of deterioration is sufficiently small and there is almost no historical change in the estimated degree of deterioration, the control changer 210 determines not to change the control, and moves the operation to step S22. In this case, it is possible to sufficiently increase the number of processed pieces by continuing the initial efficient processing conditions.

In addition, although the estimated degree of deterioration tends to rise, if there is sufficient margin before the symptom of defective processing, the control change unit 210 determines to change the control giving priority to the mold, and the operation proceeds to step S20.

On the other hand, if the estimated degree of deterioration further increases and there is not enough margin for the symptom of defective machining, the control changing unit 210 determines to change the control to give priority to machining, and the operation proceeds to step S21.

Step S20: Here, although the estimated degree of deterioration tends to rise, there is a sufficient margin for signs of processing defects. Therefore, the control change unit 210 changes the control commands (processing conditions, control parameters, etc.) of the control unit 102 in the direction of delaying mold deterioration. For example, in the process in which the workpiece 30 is fractured in the machining process (from point D to point E in FIG. 3), by weakening the pressing force and speed of the mold 20, mold wear at the time of fracture can be delayed. can.

In this way, by suppressing the increase in the estimated degree of deterioration by taking advantage of the fact that there is sufficient margin until the symptom of defective processing, it becomes possible to extend the life of the mold 20 within a reasonable range.
After this mold priority control command, the control change unit 210 shifts the operation to step S22.

Step S21: Here, the estimated degree of deterioration further increases, and there is not enough margin for signs of machining defects. Therefore, the control change unit 210 changes the control commands (machining conditions, control parameters, etc.) of the control unit 102 so as to stabilize the machining finish in preparation for the deterioration of the machining finish. For example, in a process in which the workpiece 30 is fractured during processing (from point D to point E in FIG. 3), by increasing the pressing force and speed of the mold 20, deterioration of the processing finish due to deterioration of the mold can be prevented. can be suppressed. As a result, it becomes possible to suppress defects (such as burrs) that occur in the workpiece 30 due to deterioration of the mold 20 .

At this time, the deterioration of the mold 20 progresses temporarily. As a result, the unstable period in which signs of machining defects appear or not is shortened, and it is possible to quickly move to determination of mold replacement or mold repair (steps S17 and S18).
After the processing priority control command, the control change unit 210 shifts the operation to step S22.

Step S22: The deterioration estimating unit 204 schedules (sets a timer interrupt) the next inspection time in the data acquisition unit 201 according to the history (progress) of the estimated degree of deterioration.
For example, the faster the progress of deterioration is predicted from the history of the estimated degree of deterioration, the earlier the timing of the next inspection. In addition, when the estimated degree of deterioration reaches just before the signs of machining defects, by estimating the state of all molds for each machining process, it is possible to determine mold replacement or mold repair without overlooking signs of machining defects. (Steps S17 and S18) can be performed.

The data acquisition unit 201 causes a timer interrupt at the next inspection timing, and restarts the operation from step S10, so that the next mold state estimation is performed at appropriate time intervals.
Mold state estimation is completed by the series of operations described above.

<Effect of Example 1>
(1) In the first embodiment, as shown in FIG. 5, the period from when the control internal value 102a reaches the peak value to when it reaches the end value is estimated as the time required for part of the processing. This required period is presumed to be the time from the occurrence of cracks 31 in the workpiece 30 to the completion of breakage. Since this required period synergistically reflects the double changes of "change in breaking distance" and "change in breaking speed" due to mold deterioration, it is significantly affected by mold deterioration. Therefore, mold deterioration can be estimated appropriately.

(2) In the first embodiment, phenomena related to the mold 20 (occurrence of cracks, completion of breakage) are estimated based on time-series changes in the control internal value 102a. Since the control internal value 102a is a value used for controlling the target device 100 inside the control unit 102, it can be acquired from within the target device 100 without a sensor. Therefore, a bottom dead center detection sensor and an AE (Acoustic Emission) sensor, which are required in the prior art, are no longer essential, and mold deterioration can be estimated even in a sensorless configuration.

(3) In Example 1, changes in time series of the control internal value 102a are analyzed according to a predetermined rule (peak, local minimum point) to determine the first time point t1 and the second time point t2. This predetermined rule is easy to change, and the resulting required period range can be changed immediately and freely. This level of flexibility and adaptability is not possible with conventional technology, which requires the use of dedicated sensors.

(4) If the required period is obtained based on the numerical value (absolute amount) of the control internal value 102a, the influence of low frequency noise (drift noise and 1/f noise) and offset error contained in the control internal value 102a The length of the required period varies accordingly. Therefore, there is a risk of overlooking subtle signs of mold deterioration. However, in the first embodiment, the required period is obtained from the time-series change (the relative amount of the waveform) of the control internal value 102a. Therefore, even if there is low-frequency noise or an offset error, the change in the required period is small, and the risk of overlooking signs of mold deterioration is low.

(5) In the first embodiment, two-step estimation is performed: the first estimation (estimating the required period from the time-series change of the control internal value 102a) and the second estimation (estimating the die deterioration from the required period).
If this two-stage estimation is performed in one stage (directly estimating mold deterioration from time-series changes in the control internal value 102a), there is a risk of erroneous estimation of mold deterioration because the estimation process cannot be narrowed down. .
However, since the first embodiment is a two-stage estimation, the required period is determined once in the estimation process, and then the mold deterioration is estimated. Therefore, the narrowing down of the estimation process is appropriate, and there is little risk of erroneous estimation of die deterioration.

(6) In the first embodiment, the control command current value commanded by the control unit 102 to control the drive unit 101 is used as the control internal value 102a. Since this control command current value is a value that designates an increase or decrease in the torque of the drive unit 101, it is possible to directly capture phenomena such as a delay in machining due to deterioration of the mold and an increase in the torque application period. Therefore, estimation of mold deterioration becomes short-circuited (strong correlation) and becomes easy.

(7) In the first embodiment, the state of mold deterioration is acquired, and the control command for the control unit 102 is changed according to the estimated deterioration state, thereby changing the operation of the processing using the mold 20. . Therefore, adaptive processing can be performed according to the state of deterioration of the mold 20 .

(8) In the first embodiment, a control command is issued in a direction in which the finish of processing using the mold 20 is constant according to the state of mold deterioration. Therefore, even if the mold 20 deteriorates, it is possible to avoid the occurrence of processing defects as much as possible. As a result, it becomes possible to suppress defects (such as burrs) that occur in the workpiece 30 due to deterioration of the mold 20 .

(9) In addition, in Example 1, the deterioration of the mold 20 is temporarily accelerated by making the finish of the processing treatment constant in the deteriorated mold 20 . As a result, the unstable period in which processing defects may or may not occur can be completed early, and the time to replace the mold can be shortened. As a result, there is also the effect that it is possible to suppress the occurrence of processing defects immediately before die replacement.

(10) Furthermore, in Example 1, the control command can be changed in the direction of delaying the deterioration of the mold 20 according to the deterioration state of the mold 20 . In this case, it is possible to temporarily extend the life of the mold 20 and reduce the frequency of mold replacement.

(11) Temporarily extending the life of the mold 20 in this way makes it possible to temporarily continue the operation of the target device 100 when it takes time to procure a new mold 20 . In this case, the downtime cost of the target device 100 can be reduced.

(12) In the first embodiment, as shown in FIG. 1, the mold state notification unit 205 is provided to quickly notify (the terminal of) the person in charge of maintenance of the deterioration state of the mold 20 . Therefore, it is possible to prevent situations such as a delay in replacement of the mold 20 resulting in defective processing, or sudden breakage of the mold 20 due to vibration due to deterioration or chipping.

Next, as a second embodiment, a case will be described in which the interval between the peak time t1 and the time t2 at a predetermined ratio of the peak value is set as the required period.

As shown in FIG. 9A, the period estimator 202 counts the time when the control internal value X reaches the peak value Xpk as the first time t1. Furthermore, the period estimating unit 202 counts the time when the peak value Xpk has decreased to a predetermined rate K times (the rate corresponding to point C<K<1) as a second time point t2.

The period estimation unit 202 estimates the time difference (t2-t1) between the second time point t2 and the first time point t1 as the required period (period evaluation value) required for part of the processing. The estimator 203 stores the rules for estimating the duration and degree of deterioration of the second embodiment.
Since other configurations and operations are the same as those of the first embodiment, redundant explanations are omitted here.
The required period of Example 2 reflects the time delay of the fracture phenomenon due to mold deterioration, and is therefore suitable for estimating mold deterioration.

Although the second time t2 is measured after the first time t1 in FIG. 9A, it may be a time before the first time t1 (square points in FIG. 9A). In that case, the time required for crack generation is estimated as a partial period of processing. This required period is not affected by the time delay of fracture, but is affected by the time delay of crack generation due to mold deterioration, and is suitable for estimating mold deterioration.

<Effect of Example 2>
In addition to the effects of the first embodiment described above, the second embodiment also has the following effects.

In the second embodiment, as shown in FIG. 9A, the period from when the control internal value 102a reaches a peak value to when it reaches a predetermined percentage of the peak value is estimated as the time required for part of the processing. do. This required period corresponds to part of the time from the occurrence of cracks 31 in the workpiece 30 to the completion of breakage. Since this required period also synergistically reflects the double changes of "change in breaking distance" and "change in breaking speed" due to mold deterioration, it is significantly affected by mold deterioration. Therefore, mold deterioration can be estimated appropriately.

Furthermore, in the second embodiment, the end point of the required period is set to the time when the peak value reaches a predetermined percentage. Even in cases where the value is not clear, the required period can be obtained appropriately.

Further, in the second embodiment, the period from when the control internal value 102a reaches a predetermined percentage of the peak value to when it reaches the peak value is estimated as the time required for part of the processing. This required period corresponds to part of the time from the application of force to the workpiece 30 until the crack 31 occurs. Since this required period reflects the delay in the occurrence of cracks due to mold deterioration, it is possible to appropriately estimate mold deterioration.

Furthermore, in the second embodiment, the required period and the deterioration state of the mold 20 are changed by appropriately changing the multiple K of the predetermined ratio according to the material and dimensions of the workpiece 30, the shape of the upper mold 21 and the lower mold 22, and the like. can be set so as to increase the correlation of , and the accuracy of estimating mold deterioration can be further improved.

Furthermore, as Example 3, a case will be described in which the periods before and after the peak time are set as the required period.
As shown in FIG. 9B, the period estimating unit 202 calculates a first time point t1 exceeding a predetermined ratio K times the peak value Xpk (ratio corresponding to point C<K<1) and a predetermined ratio K of the peak value Xpk. The interval at which the first point in time t2 falls below double (proportion corresponding to point C<K<1) is counted. It should be noted that the period before and after the peak point may be adjusted by varying the value of this predetermined ratio K times between the start point and the end point of the period.
The period estimation unit 202 estimates the time difference (t2-t1) between the second time point t2 and the first time point t1 as the required period (period evaluation value) required for part of the processing. The estimator 203 learns or stores the rules for estimating the duration and degree of deterioration of the third embodiment.
Since other configurations and operations are the same as those of the first embodiment, redundant explanations are omitted here.

According to the third embodiment, the time before and after the peak value Xpk is calculated as the required period. This required period is affected by both the time delay until cracks occur due to mold deterioration and the time delay before breakage. Therefore, this required period is suitable for estimating mold deterioration.

<Effect of Example 3>
In addition to the effects of the above-described embodiments, the third embodiment also has the following effects.

In Example 3, as shown in FIG. 9B, a period from when the control internal value 102a exceeds a predetermined percentage of the peak value to when it falls below a predetermined percentage of the peak value is required for part of the processing. Estimated as the required period. Since this required period includes both the delay in crack generation and the delay in breakage due to mold deterioration, it is very strongly affected by mold deterioration. Therefore, mold deterioration can be estimated more appropriately.

In the third embodiment, the correlation between the required period and the deterioration state of the mold 20 is obtained by appropriately changing the multiple K of the predetermined ratio according to the material and dimensions of the workpiece 30, the shape of the upper mold 21 and the lower mold 22, and the like. Therefore, it is possible to increase the accuracy of mold deterioration estimation.

Also, as a fourth embodiment, a case in which a plurality of required periods are used as materials for estimating the state of deterioration will be described.
The processing shown in FIG. 9C is program motion processing. Program motion machining refers to a machining method in which one machining is divided into a plurality of steps. Even if the workpiece 30 has dimensions and materials that are difficult to process in one step, it can be processed by dividing the process into a plurality of steps and performing the process step by step.

In Example 4, the first time and the second time are clocked for each step, and a plurality of required periods are calculated. Mold deterioration is estimated using the extracted required periods (t2(X1)-t1(X1) and t2(X2)-t1(X2) in FIG. 12). For this purpose, the estimator 203 learns or stores estimation rules for each of the two required periods.
Since other configurations and operations are the same as those of the first embodiment, redundant explanations are omitted here.

In Example 4, since it is possible to extract a plurality of required periods during one process, the accuracy of estimating the mold 20 is improved. Also, for example, even if an erroneous detection occurs in the required period of the first step (t2(X1)-t1(X1)), it can be detected in the second step (t2(X2)-t1(X2)). Detection can be suppressed.

<Effect of Example 4>
The fourth embodiment has the following effects in addition to the effects of the above-described embodiments.
In the fourth embodiment, as shown in FIG. 9C, one processing is divided into a plurality of steps to estimate a plurality of required periods. By combining these multiple estimates of required periods, it is possible to improve the accuracy of estimating mold deterioration. Moreover, even if one required period is unclear, the mold deterioration can be estimated only from the other required period, so that the mold deterioration can be reliably estimated.

Furthermore, as a fifth embodiment, a case will be described in which the period evaluation value is obtained based on "the length of the predetermined period" and "the magnitude of the control internal value".
For example, as shown in FIG. 9[d], the period estimator 202 obtains the result of integration by integrating the control internal value 102a over the required period estimated in the first to fourth embodiments. The period estimation unit 202 uses this integration result as the period evaluation value.
The estimator 203 learns or stores the estimation rules for the period evaluation value and the degree of deterioration of the fifth embodiment.
Since other configurations and operations are the same as those of the first embodiment, redundant explanations are omitted here.

<Effect of Example 5>
The fifth embodiment has the following effects in addition to the effects of the above-described embodiments.
In Example 5, in addition to the change in the length of the required period due to mold deterioration, the change in the magnitude of the control internal value due to mold deterioration is also taken into consideration, so that the mold deterioration can be viewed from multiple perspectives. It becomes possible to estimate That is, even if the required period is short, if the control internal value becomes excessive, the period evaluation value becomes excessive. In addition, even if the required period is short, if the control internal value becomes too small, the period evaluation value becomes too small, so it is possible to estimate without overlooking signs such as a state in which pressing is not applied due to damage to the mold 20. - 特許庁

Next, as Example 6, a case will be described in which the cloud server provides a service to the state estimation system.
FIG. 10 is a diagram for explaining management of a plurality of electric press systems using a cloud configuration.
In the figure, a cloud server 500 is connected to a plurality of electric press systems 300a, 300b, . . . via a network. The network here may be a network that can be externally connected via the Internet or the like, or a network limited in connection to the electric press systems 300a, 300b, . . . .

These electric press systems 300a, 300b, . . . are configured with target devices 100a, 100b, .
The state estimation systems 200a, 200b, .
Since other configurations and operations are the same as those of the first embodiment, redundant explanations are omitted here.

Service provision for the electric press systems 300a, 300b, . . . by the cloud server 500 will be described below.

The deterioration estimator 204 has, for example, a deterioration estimation model including the following data items and its estimation result.

"Estimation Model of Degradation"
・Type of mold 20 ・Type of control internal value 102a ・Sampling interval for data acquisition ・Definition of required period (method for determining first time point t1 and second time point t2)
・Estimation rule data for degree of deterioration and required period ・Regression line formula for degree of deterioration and required period ・Data table for regression data for degree of deterioration and required period ・Conditions for replacement (repair) of mold 20 ・Inspection for deterioration of mold 20 cycle

"Estimated result"
- Correlation coefficient between degree of deterioration and required period - History of estimated degree of deterioration - Degree of deterioration of mold 20 at the time of replacement (measured value)
・Measured value of the degree of deterioration of the mold 20 by a sensor ・Replacement cycle of the mold 20 ・Occurrence rate of processing defects

The cloud server 500 periodically collects the above-described "deterioration estimation models" and their "estimation results" from the state estimation systems 200a, 200b, .

The cloud server 500 is useful for evaluating big data (factor analysis, etc.) on similar molds 20 and target devices 100 (those with similar deterioration tendencies) and improving individual items of the "estimation results". Each item of the "degradation estimation model" is obtained.

Then, a combination of each item of the "degradation estimation model" is selected so that the evaluation function weighting the individual items of the "estimation result" in order of priority is optimized. Based on such evaluation, the cloud server 500 automatically formulates an "appropriate estimation model".

The cloud server 500 supplies the automatically formulated "appropriate estimation model" to similar state estimation systems 200a, 200b, . . . via the network. The state estimation systems 200a, 200b, . . . update the current "estimation model" to the supplied "appropriate estimation model".

On the other hand, the control change unit 210 has, for example, a control change model including the following data items and its control results.

"Control change model"
・Type of mold 20 ・Type of control internal value 102a ・Condition for selecting “mold priority” ・Control parameter for “mold priority” ・Condition for selecting “machining priority” ・Control parameter for “machining priority” ・Conditions for selecting "Do not change control" ・Control parameters for "Do not change control"

"Control result"
- Occurrence rate of machining defects - History of estimated deterioration - Degree of deterioration of mold 20 at the time of replacement (actual value)
・Measured value of deterioration degree of mold 20 by sensor ・Replacement cycle of mold 20

The cloud server 500 periodically collects the above-described "control change model" and its "control results" from the state estimation systems 200a, 200b, .

The cloud server 500 is useful for evaluating big data (factor analysis, etc.) on similar molds 20 and target devices 100 (those with similar deterioration tendencies) and improving individual items of the "control results." Obtain each item of the "control change model".

Then, a combination of each item of the "control change model" is selected so that the evaluation function obtained by weighting the individual items of the "control result" in order of priority is optimized. Based on such evaluation, the cloud server 500 automatically formulates an "appropriate control change model".

The cloud server 500 supplies the automatically formulated "appropriate control change model" to similar state estimation systems 200a, 200b, . . . via the network. The state estimation systems 200a, 200b, . . . update the current "control change model" to the supplied "appropriate control change model".

<Effect of Example 6>
In addition to the effects of the above-described embodiments, the sixth embodiment also has the following effects.
In this embodiment, in one or more electric press systems, it is possible to always maintain the estimation model and the control change model in a nearly optimal state. In addition, since the information of other electric press systems is connected to the cloud, even if the processing conditions or the processing target 30 are changed, the model is updated based on the past processing database of the other electric press system. is also possible.

<Supplementary matter of the embodiment>
In the above-described embodiment, the control command current value is used as the control internal value 102a. However, the invention is not so limited. The control internal value 102a may be control data used for the target device 100 (for example, the control unit 102 controls the driving unit 101). For example, physical quantities (force, torque, speed, acceleration, angular velocity, displacement amount, motor rotation amount, motor drive current, pulsation, load fluctuation, etc.) that change with time as the mold 20 processes the workpiece 30 are reflected. values, or values for directly or indirectly controlling them.

Further, in the above-described embodiment, as phenomena related to the processing of the target device 100, the time of crack occurrence, the time of the end value, and the time of the intermediate value before and after are estimated based on changes in the time series of the control internal value 102a. do. However, the present invention is not limited to this, and any reproducible phenomenon may be used as long as the occurrence timing changes due to wear, stain, abnormality, damage, or other deterioration of the target device 100 . Such phenomena are preferably, for example, the time of starting, the time of torque change, the point of internal division of the required period of machining, the characteristic time of machining, and the like.

Furthermore, in the above-described embodiment, changes in the time series of the control internal value 102a are analyzed by features (peaks, local minimum points, midpoints of the predetermined ratio K, etc.), and the first time point t1 and the second time point t2 are analyzed. to decide. However, the present invention is not limited to this, as long as it is an analyzable feature occurring in the time-series change of the control internal value and can be estimated as a phenomenon whose occurrence timing changes due to deterioration of the target device 100 . The characteristics of such time-series changes include points determined by inflection points, slopes, time-series changes of a predetermined pattern defined in advance, points determined by differential values and their changes, integral values (integrated values). A point determined by a value), a point determined by an envelope curve, or the like can be used as appropriate.

Furthermore, in the embodiment described above, the required period is estimated based on the chronological change in the control internal value 102a itself. However, the invention is not so limited. If the control internal value 102a is a physical quantity corresponding to the time integration of deterioration, it is preferable to estimate the required period after time-differentiating the time series change of the control internal value 102a. Further, when the control internal value 102a is a physical quantity corresponding to time differentiation of deterioration, it is preferable to estimate the required period after time-integrating the time series change of the control internal value 102a. Moreover, it is also preferable to estimate the required period after filtering the time-series change of the control internal value 102a for interference removal or feature enhancement.

In the embodiment, an example of punching using an electric press will be described. However, the present invention is suitable for processing using molds in general. For example, it is preferably applied to shearing, bending, drawing, compression, cutting, etc. using a mold. Furthermore, the present invention is suitable for use in estimating deterioration (symptoms) of consumables other than molds. For example, when an electric cutting machine is a target device, the present invention is suitable for use in monitoring signs of deterioration (change in sharpness) of a cutting blade. Further, for example, the present invention is suitable for applications such as monitoring signs of deterioration of consumables such as drill blades and end mills. Furthermore, in processing equipment, for example, the deterioration of the workpiece (defective or altered quality of the workpiece, different materials, defective formation, non-standard products, uneven weight, burrs, damage, etc.) can be determined from the required period (period evaluation value). Since it can be estimated, the present invention is suitable for use in estimating the quality of processed materials and processed products. Further, for example, the deterioration of the drive motor and the deterioration (wear and stain) of the power transmission parts in the target device can be estimated from the required period (period evaluation value), so the present invention is suitable for use in estimating signs of failure of the target device. be.

Further, in the embodiment, a sensorless configuration example will be described. However, the invention is not limited to sensorless. For example, sensors can be additionally combined for deterioration detection and combined to complement the deterioration state estimation processing of the present invention.

Further, embodiments employ correlation as an inference rule. However, the invention is not so limited. A neural network that performs machine learning using the required period (or period evaluation value) as training data and the degree of deterioration (measured value) of the target device as a teacher value, and the results of other machine learning are adopted as estimation rules. good too. If there are multiple pieces of data used for deterioration estimation, an estimation rule may be created by multivariate analysis or the like.

Reference Signs List 1 Servo motor 2 Servo amplifier 11 Eccentric shaft 12 Transmission shaft 13 Slide 20 Mold 21 Upper mold 22 Lower mold 30 Object to be processed 31 Crack 100 Target device 101 Driving unit 102 Control unit 102a Control internal value 200 State estimation system 201 Data acquisition unit 202 Period estimation unit 202a Required period 203 Estimator 204 Deterioration estimation unit 205 Mold state notification unit 210 Control change unit 300a Electric press system 500 Cloud server

Claims (12)

A state estimation system for estimating a deterioration state of a target device,
a data acquisition unit capable of acquiring control internal values used by the control unit of the target device to control the driving unit of the target device;
a period estimating unit for estimating a period required for at least a part of the drive process of the target device based on the time-series change in the control internal value, and obtaining a period evaluation value based on the required period;
an estimator that stores an estimation rule for the value of the period evaluation value and the deterioration of the target device;
A state estimation system comprising: a deterioration estimating unit that estimates deterioration of the target device by applying the period evaluation value obtained by the period estimating unit to the estimation rule of the estimator. The state estimation system of claim 1,
The period estimation unit
A state estimation system, wherein an interval between a time when the control internal value reaches a peak value and a time when the control internal value reaches an end value is estimated as the required period. The state estimation system of claim 1,
The period estimation unit
A state estimation system, wherein an interval between a point in time when the control internal value reaches a peak value and a point in time when the point reaches a predetermined percentage of the peak value is estimated as the required period. The state estimation system of claim 1,
The period estimation unit
A state estimation system, wherein an interval between a time when the control internal value exceeds a predetermined percentage of the peak value and a time when the peak value falls below the predetermined percentage is estimated as the required period. . In the state estimation system according to any one of claims 1 to 4,
The period estimation unit
A state estimation system, wherein an integration result obtained by integrating the control internal values over the estimated required period is obtained as the period evaluation value. In the state estimation system according to any one of claims 1 to 5,
The data acquisition unit
A state estimation system, wherein a control command current value commanded by the control unit to the drive unit is acquired as the control internal value. In the state estimation system according to any one of claims 1 to 6,
The period estimation unit
estimating a required period of time required for at least a part of the processing in which the target device drives the consumables based on the time-series change in the control internal value, and obtaining a period evaluation value based on the required period;
The deterioration estimator,
estimating deterioration of the consumable by applying the period evaluation value obtained by the period estimating unit to the estimation rule of the estimator;
and a control change unit that instructs the control unit to change control of the processing of the target device according to deterioration of the consumable. The state estimation system according to claim 7,
The control change unit
A state estimating system, according to deterioration of the consumable, instructing a control change in a direction in which finishing of processing using the consumable is constant. The state estimation system according to claim 7,
The control change unit
A state estimation system, wherein a control change is instructed in a direction of delaying deterioration of the consumable according to the deterioration of the consumable. A state estimation method for estimating a deterioration state of a target device,
a data acquisition step capable of acquiring a control internal value used by the control unit of the target device to control the driving unit of the target device;
a period estimating step of estimating a period required for at least a part of the drive process of the target device based on the time-series change in the control internal value, and obtaining a period evaluation value based on the required period;
a deterioration estimation step of estimating the deterioration of the target device by applying the period evaluation value obtained in the period estimation step to an estimation rule for the value of the period evaluation value and the deterioration of the target device; state estimation method. A service providing method executed by a cloud server for the state estimation system according to any one of claims 1 to 9,
an estimation model collection step of collecting and accumulating an estimation model of the deterioration of the target device and its estimation results from one or more of the state estimation systems via a network;
an estimation model optimization step of evaluating the accumulated estimation model based on the estimation result and optimizing the estimation model;
and an estimation model update step of updating the estimation model of the state estimation system based on the optimized estimation model. In the service providing method according to claim 11,
a control change model collection step of collecting and accumulating a control change model of the control change unit and its control result from the state estimation system including the control change unit via a network;
a control change model optimization step of evaluating the accumulated control change model based on the control result and optimizing the control change model;
and a control change model update step of updating the control change model of the state estimation system based on the optimized control change model.

Images