Industrial robot and monitoring method for industrial robot

Abstract

To relatively easily grasp abnormality even in an abnormal mode of a mode difficult to be detected by time series data of a current value of a servo motor.SOLUTION: The servo control unit 21 of the control device 20 controls driving of the servo motor of each axis based on a control command from the calculation unit 22. The current value acquisition unit 31 of the monitoring device 30 acquires the time-series data It of the command current value from the control device 20. The axis value acquisition unit 32 acquires the time-series data Pt of the axis value of each servo motor. The first waveform acquisition unit 33 acquires a first waveform Wpi on the basis of the time-series data It and the time-series data Pt. The first waveform Wpi is a waveform indicating a relationship between the axis value and the current value of the servo motor. The display control unit 34 displays the first waveform Wpi acquired by the first waveform acquisition unit 33 on the display device 40. In first waveform Wpi, an abnormality in the drive unit can be relatively easily ascertained compared to the current waveform.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to an industrial robot and a method for monitoring an industrial robot. [Background technology]

[0002] Japanese Patent Application Laid-Open Publication No. 2019-67240 (Patent Document 1) discloses a method for estimating an abnormality in a robot that uses a servo motor to drive an arm. In Patent Document 1, the observed value is the current value flowing through the servo motor or the position deviation of the servo motor. In Patent Document 1, an abnormality is detected using the mean square value or peak value of the time-series data of the current value. Patent Document 1 also states that an abnormality may be detected using the mean square value or peak value of the time-series data of the position deviation. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2019-67240 Summary of the Invention [Problem to be solved by the invention]

[0004] In Patent Document 1, time series data of servo motor current values ​​or time series data of position deviations are used as the observed values ​​for abnormality detection. However, when time series data of current values ​​or time series data of position deviations are used as the observed values, depending on the type of abnormality, it may be difficult to distinguish between normal and abnormal, and the abnormality may not be detected.

[0005] An object of the present disclosure is to make it relatively easy to grasp abnormalities even if the abnormalities are of a type that is difficult to detect using time-series data of servo motor current values. [Means for solving the problem]

[0006] The industrial robot of the present disclosure is an industrial robot driven by a drive unit including a servo motor and a reducer. The industrial robot includes a control device that controls the servo motor and a monitoring device that monitors the drive unit. The monitoring device includes a first waveform acquisition unit that acquires a first waveform that represents the relationship between the shaft position of the servo motor and the current value of the servo motor.

[0007] According to this configuration, the first waveform acquisition unit of the monitoring device acquires a first waveform that is the relationship between the shaft position of the servo motor and the current value of the servo motor. The shaft position is the rotation angle of the servo motor, and is also referred to as the "shaft value" in this disclosure. The current value is the value of the current flowing through the servo motor. The first waveform is, for example, a waveform of the current value versus the shaft value, with the vertical axis representing the current value and the horizontal axis representing the shaft value.

[0008] For example, in an abnormal mode where there is an abnormality in the reducer, it is expected that significant characteristics will appear during acceleration, deceleration, and feedback control, compared to when the servo motor is operating at a constant speed. However, time-series data of the servo motor's current value (also called "current waveform") may not show a significant difference between the current waveform in normal and abnormal conditions. Even in such cases, the first waveform, which is the relationship between the axis value and the current value, is likely to show a significant difference between normal and abnormal conditions, making it relatively easy to identify the abnormality.

[0009] Preferably, the signal processing device may further include a filter section that passes a predetermined frequency component of the first waveform.

[0010] Depending on the type of abnormality, characteristics may appear more prominently in a predetermined frequency component of the first waveform. With this configuration, abnormalities can be relatively easily identified in various abnormality types (abnormal modes) by the waveform output from the filter unit that passes the predetermined frequency component of the first waveform. The predetermined frequency component may be in the high-frequency region of the first waveform or in the low-frequency region of the first waveform. Furthermore, the predetermined frequency component may be in a specific frequency band of the first waveform.

[0011] Preferably, the first waveform acquisition section may acquire the first waveform based on time series data of a current value flowing through the servo motor and time series data of a shaft position of the servo motor.

[0012] The current value may be, for example, a command value (command current value) of a control device. The axis position (axis value) may be, for example, a value detected by an encoder (rotary encoder) of a servo motor. The first waveform is acquired by associating the current value and axis value at the same time in the time-series data of the current value and the time-series data of the axis value.

[0013] Preferably, the monitoring device may further include an abnormality detection section that detects the degree of an abnormality that has occurred in the drive unit based on the first waveform.

[0014] According to this configuration, the abnormality detection unit detects the degree of the abnormality that has occurred in the drive unit based on the first waveform. As a result, when an abnormality occurs in the drive unit, the abnormality detection unit can detect the degree of the abnormality.

[0015] Preferably, the degree of abnormality detected by the abnormality detection section may be the degree of deviation between the first waveform when the drive unit is operating normally and the first waveform acquired by the first waveform acquisition section.

[0016] According to this configuration, the degree of abnormality can be detected based on the degree of deviation between the first waveform acquired by the first waveform acquisition section and the first waveform when the drive unit is operating normally.

[0017] Preferably, the abnormality detection unit may detect the degree of the abnormality using a predetermined frequency component of the first waveform.

[0018] Depending on the type of abnormality, the characteristics may be more pronounced in a predetermined frequency component of the first waveform. With this configuration, the degree of abnormality is detected using the predetermined frequency component of the first waveform, making it possible to relatively easily detect abnormalities in various abnormality types (abnormal modes). The predetermined frequency component may be in the high-frequency region of the first waveform or in the low-frequency region of the first waveform. The predetermined frequency component may also be in a specific frequency band of the first waveform.

[0019] The present disclosure provides a method for monitoring an industrial robot driven by a drive unit including a servo motor and a reducer, and includes a first step of acquiring a first waveform representing the relationship between the shaft position of the servo motor and the current value of the servo motor, and a second step of detecting the degree of an abnormality occurring in the drive unit based on the first waveform.

[0020] According to this method, a first waveform, which is the relationship between the shaft position of the servo motor and the current value of the servo motor, is acquired, and the degree of abnormality that has occurred in the drive unit is detected based on the first waveform. The first waveform, which is the relationship between the shaft position (shaft value) and the current value, tends to show a significant difference between normal and abnormal conditions, making it possible to detect abnormalities relatively easily.

[0021] Preferably, the second step may detect the degree of abnormality using a trained model that has learned the first waveform when the drive unit is operating normally and the first waveform obtained in the first step.

[0022] According to this method, the degree of abnormality is detected using a trained model that has learned the first waveform when the drive unit is operating normally, so the monitoring method can be deployed in a general-purpose manner. [Effects of the Invention]

[0023] According to the present disclosure, even if an abnormality is difficult to detect using time-series data of current values, it can be relatively easily detected. [Brief explanation of the drawings]

[0024] [Figure 1] 1 is a diagram showing a schematic configuration of an industrial robot according to an embodiment of the present invention; [Figure 2] FIG. 2 is a diagram illustrating an example of functional blocks of a control device and a monitoring device according to the present embodiment. [Figure 3] 10 is a flowchart showing an example of a first waveform display control executed by the monitoring device. [Figure 4] 10(A) and 10(B) are diagrams showing waveforms (current waveforms) of time-series data It of current values. [Figure 5] 10(A) and 10(B) are diagrams showing the first waveform Wpi. [Figure 6] FIG. 10 is a diagram illustrating an example of functional blocks of a control device and a monitoring device according to a second embodiment. [Figure 7] 10 is a flowchart illustrating an example of abnormality display control executed by the monitoring device. [Figure 8] 10A and 10B are diagrams illustrating the transition of the degree of deviation. [Figure 9] FIG. 11 is a diagram illustrating an example of functional blocks of a control device and a monitoring device according to a third embodiment. [Figure 10] 11(A) and 11(B) are diagrams showing an example of a first waveform Wpi displayed on a display device in the third embodiment. [Figure 11] FIG. 13 is a diagram illustrating an example of functional blocks of a control device and a monitoring device according to a fourth embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0025] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the following description, the same or corresponding parts in the drawings will be denoted by the same reference numerals, and description thereof may not be repeated.

[0026] (Embodiment 1) FIG. 1 is a diagram showing a schematic configuration of an industrial robot 100 according to this embodiment. In FIG. 1, the industrial robot 100 includes a six-axis vertical articulated robot 10, a control device 20, a monitoring device 30, and a display device 40. The six-axis vertical articulated robot (hereinafter also referred to as the "robot main body") 10 is a known six-axis robot that is installed on a base 10A and includes a first axis 11, a second axis 12, a third axis 13, a fourth axis 14, a fifth axis 15, and a sixth axis 16. Each of the first axis 11 to the sixth axis 16 is provided with a drive unit (not shown) including a servo motor and a reducer. Each of the first axis 11 to the sixth axis 16 (each joint) is driven by a corresponding servo motor (drive unit).

[0027] The control device 20 controls the servo motors of each axis and controls the operation of the robot main body 10. The monitoring device 30 monitors the status of the drive units of each axis. The control device 20 and monitoring device 30 may be composed of a CPU (Central Processing Unit) that performs various processes, memories including ROM (Read Only Memory) that stores programs and data and RAM (Random Access Memory) that stores CPU processing results, and input / output ports for exchanging information with the outside. The encoder 18 detects the axis position (axis value) of the servo motor of each axis. The display device 40 is a display that displays information sent from the monitoring device 30.

[0028] FIG. 2 is a diagram showing an example of functional blocks of the control device 20 and the monitoring device 30 in this embodiment. Referring to FIG. 2, the control device 20 includes a servo control unit 21 and a calculation unit 22. The servo control unit 21 controls the driving of the servo motors of each axis based on control commands from the calculation unit 22. The calculation unit 22 generates control commands based on a program taught offline or online, and outputs the control commands to the servo control unit 21. The control commands include, for example, the axis values ​​and target speed values ​​of each servo motor. The servo control unit 21 uses the axis values ​​detected by the encoder 18 to generate command current values ​​to flow through each servo motor so that the axis values ​​and speed become the target values, and performs feedback control of each servo motor.

[0029] The monitoring device 30 includes a current value acquiring unit 31, an axis value acquiring unit 32, a first waveform acquiring unit 33, and a display control unit 34. The current value acquiring unit 31 acquires time series data It of command current values ​​from the control device 20. The time series data It is time series data of the current values ​​flowing through each servo motor. The axis value acquiring unit 32 acquires time series data Pt of the axis values ​​of each servo motor. The time series data Pt may be time series data of the axis values ​​detected by the encoder 18.

[0030] The first waveform acquisition unit 33 acquires a first waveform Wpi based on the time series data It and the time series data Pt. The first waveform Wpi is a waveform showing the relationship between the axis value and the current value of the servo motor. In this embodiment, the first waveform Wpi is a waveform of the current value versus the axis value, with the axis value on the horizontal axis and the current value on the vertical axis. The first waveform acquisition unit 33 acquires the first waveform Wpi by plotting the current value and the axis value at the same time on a plane where the axis value is on the horizontal axis and the current value is on the vertical axis, using the time labels (time information) of the time series data It and the time series data Pt. The display control unit 34 displays the first waveform Wpi acquired by the first waveform acquisition unit 33 on the display device 40. In the present disclosure, the first waveform Wpi may also be referred to as "axis value current data."

[0031] 3 is a flowchart showing an example of first waveform display control executed by the monitoring device 30. This flowchart is processed at predetermined intervals while the industrial robot 100 is in operation. The flowchart is also processed for each of the servo motors of the first axis 11 to the sixth axis 16. In step (hereinafter, step will be abbreviated as "S") 10, time series data It of the current value (command current value) and time series data Pt of the axis value are acquired from the control device 20. The time series data It and the time series data Pt are data for the same time range, and may be data for, for example, several seconds to several tens of seconds.

[0032] In S11, the first waveform Wpi is calculated by plotting the current value and axis value at the same time using the time labels (time information) of the time series data It and the time series data Pt on a plane with the axis value on the horizontal axis and the current value on the vertical axis. In S12, the first waveform Wpi calculated in S11 is displayed on the display device 40, and the current routine is terminated.

[0033] FIG. 4 shows the waveform of time-series data It of current values ​​(current waveform). FIG. 4(A) shows the current waveform of the servo motor of the third axis 13, and FIG. 4(B) shows the current waveform of the servo motor of the second axis 12. In FIG. 4, the vertical axis represents current value, and the horizontal axis represents time. The solid line shows the current waveform when the drive unit is normal. The dashed line shows the current waveform when the drive unit is abnormal. In FIG. 4, an abnormality in the drive unit occurs when some of the teeth of the gears that make up the reducer are worn. As shown in FIGS. 4(A) and (B), the servo motor current waveform has many overlapping regions between the normal (solid line) and abnormal (dashed line) waveforms, and the waveform shapes are also similar. For this reason, it is difficult to determine an abnormality in the drive unit from the servo motor current waveform.

[0034] FIG. 5 shows the first waveform Wpi. FIG. 5(A) shows the first waveform Wpi of the servo motor of the third axis 13, and FIG. 5(B) shows the first waveform Wpi of the servo motor of the second axis 12. In FIG. 5, the vertical axis represents the current value, and the horizontal axis represents the axis value. The solid line shows the first waveform Wpi when the drive unit is normal. The dashed line shows the first waveform Wpi when the drive unit is abnormal. As in the example of FIG. 4, an abnormality in the drive unit occurs when some of the tips of the teeth of the gears that make up the reducer are worn. The first waveform Wpi in FIGS. 5(A) and (B) is calculated based on the current waveforms (time-series data It) in FIGS. 4(A) and (B). Comparing the first waveform Wpi under normal conditions (solid line) and under abnormal conditions (dashed line), the peak value, peak-to-peak value, and other characteristics are easily recognized in the areas enclosed by the dashed lines in FIGS. 5(A) and (B). For this reason, it is relatively easier to detect an abnormality in the drive unit from the first waveform Wpi than from the current waveform (see FIG. 4).

[0035] The display control unit 34 (corresponding to the process of S12) may display the first waveform Wpi on the display device 40 together with the identification information of the drive units (servo motors) of the first axis 11 to the sixth axis 16 each time the first waveform Wpi is calculated in S11. A supervisor (worker) of the industrial robot 100 may select one of the first axis 11 to the sixth axis 16, and the first waveform Wpi corresponding to the selected axis (joint) may be displayed on the display device 40. The display device 40 may display multiple first waveforms Wpi for each axis (for example, the most recent 10 waveforms) in an overlapping manner. The display device 40 may also display all of the first waveforms Wpi for the first axis 11 to the sixth axis 16. A supervisor (worker) of the industrial robot 100 can relatively easily detect an abnormality in the drive unit of the robot body 10 by observing the first waveform Wpi displayed on the display device 40.

[0036] (Embodiment 2) In the second embodiment, the monitoring device 30A includes an abnormality detection unit that detects the degree of an abnormality that has occurred in the drive unit based on the first waveform Wpi. The general configuration of the industrial robot of the second embodiment is similar to that of the industrial robot 100 of the first embodiment, and therefore a description thereof will be omitted. Figure 6 is a diagram showing an example of functional blocks of the control device 20 and the monitoring device 30A in the second embodiment. In the second embodiment, the control device 20 is similar to that of the first embodiment, and therefore a description thereof will be omitted.

[0037] Referring to FIG. 6, in the monitoring device 30A, the current value acquiring unit 31, the axis value acquiring unit 32, and the first waveform acquiring unit 33 are the same as those in the first embodiment. The abnormality detecting unit 35 detects the degree of abnormality that has occurred in the drive unit based on the first waveform Wpi acquired by the first waveform acquiring unit 33. In this embodiment, the abnormality detecting unit 35 uses a trained model 36 that has learned the first waveform Wpi in a normal state to determine the deviation D of the first waveform Wpi acquired by the first waveform acquiring unit 33. The trained model 36 is a trained model (normal model) that is generated by learning data (normal data) when the drive unit is operating normally through unsupervised learning. The trained model 36 is generated for each drive unit of each axis.

[0038] The trained model 36 may be a trained model generated by using, for example, a VAE (Variational Autoencoder) to train the first waveform Wpi when the drive unit is operating normally. In this case, the abnormality detection unit 35 inputs the first waveform Wpi acquired by the first waveform acquisition unit 33 into the trained model 36 (VAE) to obtain an output X. Then, the abnormality detection unit 35 finds the difference between the input first waveform Wpi and the output X to calculate the deviation D.

[0039] If the input first waveform Wpi is normal (when the normal first waveform Wpi is input to the trained model 36), the difference between the first waveform Wpi and the output X is small. In this case, the deviation D is 0 or a small value. If the input first waveform Wpi is abnormal (when the abnormal first waveform Wpi is input to the trained model 36), a large difference occurs between the first waveform Wpi and the output X. The greater the degree of deviation between the normal first waveform Wpi and the input (abnormal) first waveform Wpi, the larger the difference. The abnormality detection unit 35 calculates the deviation D so that the larger this difference is, the larger the deviation D becomes. The deviation D is calculated for each drive unit of each axis.

[0040] If the deviation D detected by the abnormality detection unit 35 is greater than a threshold value, the abnormality determination unit 37 determines that the corresponding drive unit is abnormal. For example, the abnormality determination unit 37 may calculate an average value Dav of the deviation D over a set period, and determine that the drive unit is abnormal if the average value Dav is equal to or greater than a predetermined threshold value α.

[0041] When the abnormality determination unit 37 determines that an abnormality has occurred in the drive unit, the display control unit 34 displays, on the display device 40, a message to the effect that an abnormality has occurred and information about the axis (joint) in which the abnormality has occurred. At this time, information about the deviation degree D may also be displayed on the display device 40.

[0042] FIG. 7 is a flowchart showing an example of anomaly display control executed by the monitoring device 30A. This flowchart is processed at predetermined intervals while the industrial robot 100 is in operation. The processing is also performed for each of the servo motors of the first axis 11 to the sixth axis 16. S20 and S21 are the same as S10 and S11 in the first embodiment. In S22, a deviation D between the first waveform Wpi calculated in S21 and the trained model (normal model) is calculated. The deviation D may be calculated from the difference between the first waveform Wpi and the output X obtained by inputting the first waveform Wpi to the trained model (VAE), as described above.

[0043] In the next step S23, the deviation D calculated in step S22 is stored in memory. In step S24, it is determined whether a set period has elapsed. The set period may be, for example, the operating time of the industrial robot 100 in one day. Alternatively, the set period may be a time period set in advance (for example, four hours). If the set period has elapsed, a positive determination is made, and the process proceeds to step S25. If the set period has not elapsed, a negative determination is made, and the current routine is terminated. As a result, in step S23, the deviation D calculated during the set period is stored in memory.

[0044] In S25, the average value Dav of the deviations D stored in memory is calculated. The average value Dav may be a simple average (arithmetic mean) of the deviations D accumulated in memory over a set period. In S26, it is determined whether the average value Dav is equal to or greater than a threshold value α. If the average value Dav is equal to or greater than the threshold value α, the process proceeds to S27. If the average value Dav is less than the threshold value α, the process proceeds to S28.

[0045] In S27, it is determined that an abnormality has occurred in the drive unit, and the current routine is terminated. In this case, the fact that an abnormality has occurred, information about the axis (joint) where the abnormality has occurred, and the average value Dav may be displayed on the display device 40. In S28, it is determined that the drive unit is normal, and the current routine is terminated.

[0046] Fig. 8 is a diagram illustrating the transition of the deviation. Fig. 8(A) shows the transition of the average value Dav of the deviation D calculated using the axial value current data (first waveform Wpi) (the transition of the average value Dav obtained in the flowchart of the abnormality display control in Fig. 7). Fig. 8(B) shows the transition of the average value Fav of the deviation F calculated using the time-series data It of the current value (current waveform).

[0047] The deviation F (average value Fav) is calculated in the same way as the deviation D (average value Dav) by using a trained model (normal model) generated by unsupervised learning of the time series data It when the drive unit is operating normally.

[0048] In Fig. 8(A), the average value Dav is the average value of the deviation D during one day's operation time of the industrial robot 100. In Fig. 8(B), the average value Fav is the average value of the deviation F during one day's operation time of the industrial robot 100. In Figs. 8(A) and (B), the horizontal axis represents the date.

[0049] 8(A) and 8(B), the robot body 10 is operated using a normal drive unit from the first day to day A. After operation on day A ends and before operation on day B, the day after day A, the drive unit is replaced with a drive unit in which some of the tips of the teeth of the gears that make up the reducer are worn (hereinafter also referred to as an "abnormal drive unit"). From day B to day C, the robot body 10 is operated using the abnormal drive unit. From day C ends and before operation on day D, the day after day C, the drive unit is replaced with a normal drive unit. From day D to day E, the robot body 10 is operated using the normal drive unit. Using a similar method, the robot body 10 is operated using the abnormal drive unit from day F (the day after day E) to day G, and from day H (the day after day G) onwards, the robot body 10 is operated using the normal drive unit.

[0050] As shown in FIG. 8(A), the average value Dav of the deviation D calculated using the first waveform Wpi (axial current data) differs significantly between normal (when operating with a normal drive unit) and abnormal (when operating with an abnormal drive unit), making it possible to clearly distinguish between normal and abnormal conditions. As shown in FIG. 8(B), the average value Fav of the deviation F calculated using the current waveform (current time-series data) may not clearly distinguish between normal and abnormal conditions. Therefore, by using the deviation D (average value Dav), it is possible to relatively easily detect an abnormality in the drive unit of the robot main body 10. Furthermore, by using the deviation D (average value Dav), it is easy to set a threshold value for determining an abnormality, thereby reducing false detection of an abnormality in the drive unit. The deviation D (average value Dav) corresponds to an example of the "degree of abnormality" in this disclosure.

[0051] (Embodiment 3) In the third embodiment, a low-pass filter and a high-pass filter are added to the monitoring device 30 in the industrial robot 100 of the first embodiment. Fig. 9 is a diagram showing an example of functional blocks of the control device 20 and the monitoring device 30B in the third embodiment. In the third embodiment, the control device 20 is the same as in the first embodiment, and therefore a description thereof will be omitted.

[0052] 9, in monitoring device 30B, current value acquiring unit 31, axis value acquiring unit 32, and first waveform acquiring unit 33 are the same as those in embodiment 1. Low-pass filter 38 passes frequency components lower than a cutoff frequency Fc1 of first waveform Wpi acquired by first waveform acquiring unit 33, and blocks (attenuates) frequency components higher than cutoff frequency Fc1. High-pass filter 39 passes frequency components higher than a cutoff frequency Fc2 of first waveform Wpi, and blocks (attenuates) frequency components lower than cutoff frequency Fc2.

[0053] The display control unit 34 displays the low-frequency components of the first waveform Wpi output from the low-pass filter 38 on the display device 40. The display control unit 34 displays the high-frequency components of the first waveform Wpi output from the high-pass filter 39 on the display device 40.

[0054] 10A and 10B show an example of the first waveform Wpi displayed on the display device 40 in the third embodiment. FIG. 10A shows the low-frequency components of the first waveform Wpi output from the low-pass filter 38 (the low-frequency region of the first waveform Wpi), and FIG. 10B shows the high-frequency components of the first waveform Wpi output from the high-pass filter 39 (the high-frequency region of the first waveform Wpi). The solid line shows the waveform when the drive unit is normal, and the dashed line shows the waveform when the drive unit is abnormal. In this example, the cutoff frequency Fc1 of the low-pass filter 38 and the cutoff frequency Fc2 of the high-pass filter 39 are the same frequency.

[0055] Comparing the waveforms in Figures 10(A) and 10(B), in this abnormal mode, the low-frequency components of the first waveform Wpi clearly distinguish between normal and abnormal conditions, but the high-frequency components of the first waveform Wpi do not. Therefore, the waveform output from the filter section that passes a predetermined frequency component of the first waveform Wpi can relatively easily identify abnormalities in the drive unit in various abnormal conditions (abnormal modes). Furthermore, using the waveform output from the filter section that passes a predetermined frequency component of the first waveform Wpi may also make it possible to identify the abnormal mode of the drive unit. Note that if it is sufficient to determine whether a specific abnormal mode is normal or abnormal, and a frequency band in which the characteristics of normal and abnormal conditions are clearly identified in advance, only the waveform of that frequency band may be displayed on the display device 40.

[0056] (Fourth embodiment) In the fourth embodiment, a low-pass filter and a high-pass filter are added to the monitoring device 30A in the industrial robot of the second embodiment. Fig. 11 is a diagram showing an example of functional blocks of the control device 20 and the monitoring device 30C in the fourth embodiment. In the third embodiment, the control device 20 is the same as in the first embodiment, and therefore a description thereof will be omitted.

[0057] 11, in monitoring device 30C, current value acquiring unit 31, axis value acquiring unit 32, and first waveform acquiring unit 33 are the same as in embodiment 1. As in embodiment 3, low-pass filter 38 passes only frequency components lower than cutoff frequency Fc1 of first waveform Wpi acquired by first waveform acquiring unit 33, and high-pass filter 39 passes only frequency components higher than cutoff frequency Fc2 of first waveform Wpi.

[0058] The anomaly detection unit 35 detects the degree of an anomaly occurring in the drive unit based on the low-frequency components (low-frequency region of the first waveform Wpi) of the first waveform Wpi output from the low-pass filter 38 and the high-frequency components (high-frequency region of the first waveform Wpi) of the first waveform Wpi output from the high-pass filter 39. In this embodiment, a trained model (also referred to as normal model A) that has trained the low-frequency components of the first waveform Wpi in the normal state and a trained model (also referred to as normal model B) that has trained the high-frequency components of the first waveform Wpi in the normal state are set as the trained model 36. As in the second embodiment, the anomaly detection unit 35 inputs the low-frequency components of the first waveform Wpi output from the low-pass filter 38 to the normal model A (VAE) and calculates the degree of deviation from the difference between the obtained output X and the low-frequency components of the input first waveform Wpi. Furthermore, the abnormality detection unit 35 inputs the high-frequency components of the first waveform Wpi output from the high-pass filter 39 to the normal model B (VAE), and calculates the deviation from the difference between the obtained output X and the high-frequency components of the input first waveform Wpi.

[0059] The abnormality determination unit 37 determines that the corresponding drive unit is abnormal if the deviation in the low-frequency component of the first waveform Wpi is greater than a threshold value, or if the deviation in the high-frequency component of the first waveform Wpi is greater than a threshold value. When the abnormality determination unit 37 determines that an abnormality has occurred in the drive unit, the display control unit 34 displays on the display device 40 a message that an abnormality has occurred and information about the axis (joint) in which the abnormality has occurred. At this time, the display device 40 may also display in which frequency band the abnormality has been detected.

[0060] According to the fourth embodiment, abnormalities in the drive unit can be detected relatively easily in various abnormal conditions (abnormal modes) based on the waveform output from the filter unit that passes predetermined frequency components of the first waveform Wpi. Furthermore, by determining in which frequency band the abnormality is detected, it becomes possible to identify the abnormal mode of the drive unit. Note that if it is sufficient to determine whether a specific abnormal mode is normal or abnormal and a frequency band in which characteristics of normal and abnormal states are clearly evident is known in advance, the abnormality detection unit 35, trained model 36, and abnormality determination unit 37 may perform the above-described processing only on waveforms in that frequency band.

[0061] In the above embodiment, the command current value of the control device 20 is used as the time-series data It of the current value. However, the current flowing through the servo motor may be detected using a current sensor to obtain the time-series data It. In this case, if the servo motor is a three-phase motor, the time-series data It may be obtained from the current values ​​of the U, V, and W phases using a predetermined calculation formula.

[0062] In the above-described second and fourth embodiments, the deviation D is calculated using unsupervised learning using a VAE. However, any method for calculating the deviation D may be used as long as it can detect the difference between the first waveform Wpi in a normal state and the first waveform Wpi in an abnormal state. For example, unsupervised learning using a GAN (Generative Adversarial Network) may be used. Furthermore, although the average value Dav of the deviation D is compared with a threshold value to determine whether or not there is an abnormality, the determination may be made using the median Dc instead of the average value Dav, or other statistics may be used.

[0063] In the above-described third and fourth embodiments, the first waveform Wpi is divided into a low-frequency region and a high-frequency region using the low-pass filter 38 and the high-pass filter 39. However, either the low-pass filter 38 or the high-pass filter 39 may be used alone. Alternatively, a predetermined frequency band of the first waveform Wpi may be extracted using a band-pass filter. In this case, multiple band-pass filters may be provided to extract components of multiple frequency bands from the first waveform Wpi.

[0064] In the above-described third and fourth embodiments, the cutoff frequencies Fc1 and Fc2 may be set based on the knowledge (experience) of the supervisor (operator). Furthermore, the frequency band to be extracted may be extracted by performing principal component analysis (PCA) on the frequency space of the first waveform Wpi using a frequency spectrum divided into multiple bands as explanatory variables, and extracting a frequency band that is the sum of principal components whose contribution rate is equal to or greater than a certain value. For example, the frequency band with a high contribution rate may be only the first principal component, or may be the sum of the first to second principal components.

[0065] In the above embodiment, an industrial robot 100 including a six-axis vertical articulated robot 10 has been described, but the robot may have one or two axes instead of six axes. In this disclosure, "industrial robot" is not limited to robots used in secondary industries, but also includes robots used in primary and tertiary industries.

[0066] The embodiments disclosed herein should be considered to be illustrative in all respects and not restrictive. The scope of the present disclosure is defined by the claims, not by the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0067] 10 6-axis vertical articulated robot, 10A base, 11 first axis, 12 second axis, 13 third axis, 14 fourth axis, 15 fifth axis, 16 sixth axis, 18 encoder, 20 control device, 21 servo control unit, 22 calculation unit, 30, 30A, 30B, 30C monitoring device, 31 current value acquisition unit, 32 axis value acquisition unit, 33 first waveform acquisition unit, 34 display control unit, 35 abnormality detection unit, 36 trained model, 37 abnormality judgment unit, 38 low-pass filter, 39 high-pass filter, 40 display device, 100 industrial robot.

Claims

1. An industrial robot driven by a drive unit including a servo motor and a reducer, a control device for controlling the servo motor; a monitoring device that monitors the drive unit, The monitoring device an industrial robot including a first waveform acquisition unit that acquires a first waveform that is a relationship between an axial position of the servo motor and a current value of the servo motor.

2. The monitoring device 2. The industrial robot according to claim 1, further comprising a filter section that passes a predetermined frequency component of the first waveform.

3. 2. The industrial robot according to claim 1, wherein the first waveform acquisition unit acquires the first waveform based on time series data of a current value flowing through the servo motor and time series data of a shaft position of the servo motor.

4. The monitoring device The industrial robot according to claim 1 , further comprising an abnormality detection unit that detects the degree of an abnormality that has occurred in the drive unit based on the first waveform.

5. 5. The industrial robot according to claim 4, wherein the degree of abnormality is a degree of deviation between the first waveform when the drive unit is operating normally and the first waveform acquired by the first waveform acquisition unit.

6. The industrial robot according to claim 4 , wherein the abnormality detection unit detects the degree of the abnormality by using a predetermined frequency component of the first waveform.

7. A method for monitoring an industrial robot driven by a drive unit including a servo motor and a reducer, comprising: a first step of acquiring a first waveform representing a relationship between a shaft position of the servo motor and a current value of the servo motor; a second step of detecting the degree of an abnormality occurring in the drive unit based on the first waveform.

8. 8. The industrial robot monitoring method according to claim 7, wherein the second step detects the degree of the abnormality using a trained model that has learned the first waveform when the drive unit is operating normally and the first waveform acquired in the first step.

Images