JP2020153930A - Abnormality determination device and abnormality determination method - Google Patents

Abstract

To suppress an erroneous detection of an abnormality at a work apparatus.SOLUTION: An abnormality determination device 1 includes an abnormality detection part 13 and a result correction part 15. The abnormality detection part 13 is configured to detect an abnormality of each of a plurality of robots 3A-3D according to sensor data obtained from the plurality of robots 3A-3D. When the abnormality detection part 13 detects an abnormality of any of the plurality of robots 3A-3D, the result correction part 15 corrects a detection result if there is a correlation between an abnormality detection state of a robot 3A in which the abnormality is detected and abnormality detection states of robots 3B-3D other than the robot 3A.SELECTED DRAWING: Figure 1

Description

The present invention relates to a technique for determining an abnormality in a working device.

In Patent Document 1, when detecting an abnormality of a plurality of motors, data such as the number of rotations of each motor is detected, and based on the correlation between the motors, monitoring of the motor that determines that the data outside the predetermined range is abnormal. A method has been proposed.

JP-A-2017-32567

In product manufacturing in factories, so many false positives occur that we try to eliminate the oversight of abnormalities in work robots. If a large number of false positives occur, the efficiency of product manufacturing will decrease. When an abnormality is detected, it is necessary to determine whether the abnormality is caused by the work robot itself, the entire factory, the manufacturing line, or the work object (work) of the work robot.

Patent Document 1 is a technique for monitoring an abnormality of a plurality of motors included in a transport device. However, when the technique of Patent Document 1 is applied to a work robot, the motor for operating the work robot does not always rotate like the motor provided in the transfer device, but for obtaining the correlation between the motors. There was a problem that the number of data was small. Further, Patent Document 1 does not consider suppressing false detection.

The present invention has been made in view of the above, and an object of the present invention is to suppress erroneous detection of abnormalities in working equipment.

The abnormality determination device according to one aspect of the present invention determines an abnormality of each of a plurality of work devices based on sensor data acquired from each of the plurality of work devices. The abnormality determination device corrects the determination result of the first work device based on the abnormality detection state of the first work device determined to be abnormal and the abnormality detection state of the second work device other than the first work device. ..

According to the present invention, it is possible to suppress erroneous detection of an abnormality in a working device.

It is a functional block diagram which shows the structure of the abnormality determination apparatus of 1st Embodiment. It is a flowchart which shows the process flow of the abnormality determination method of 1st Embodiment. It is a figure which shows the example which there is no correlation with the degree of abnormality of two robots. It is a figure which shows the example which correlates with the degree of abnormality of two robots. It is a functional block diagram which shows the structure of the abnormality determination apparatus of 2nd Embodiment. It is a figure for demonstrating the robot of the same process. It is a flowchart which shows the process flow of the abnormality determination method of 2nd Embodiment. It is a flowchart which shows the process flow of the abnormality determination method of 3rd Embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the description of the drawings, the same parts are designated by the same reference numerals and the description thereof will be omitted.

[First Embodiment]
The abnormality determination device of the first embodiment will be described with reference to FIG. The abnormality determination device 1 shown in FIG. 1 receives sensor data from sensors 31A to 31D attached to each of the plurality of robots 3A to 3D, and determines an abnormality of each of the plurality of robots 3A to 3D based on the sensor data. It is a device to do.

Robots 3A to 3D are work devices that drive a rotation mechanism to operate an arm portion. For example, the robots 3A to 3D are equipment for performing production at a production site, or multi-axis joint type robots of the same type that form a part of the equipment. Although not shown, the robots 3A to 3D include a plurality of operating axes as a rotation mechanism. Each operating shaft is provided with a motor and a speed reducer as a driving unit for driving the operating shaft. By driving the motor, the operation shaft is operated via the speed reducer to control the operation of the robot arm. A work hand, a spot welder, and the like are attached to the robot arm. Each of the robots 3A to 3D receives a control signal according to the work process from the robot motion control device (not shown), and switches the operation speed, the operation angle, the work order, and the work contents, etc. Do.

The sensors 31A to 31D are sensors installed in or near the driving unit of the robots 3A to 3D in order to quantitatively detect the state of the robots 3A to 3D. For example, the sensors 31A to 31D are an optical sensor, a sound sensor, an acceleration sensor, a vibration sensor, a strain sensor, an acoustic emission sensor, and a temperature sensor. The sensors 31A to 31D may be cameras that capture images or moving images.

The abnormality determination device 1 includes a signal processing unit 11, a storage unit 12, an abnormality detection unit 13, a correlation detection unit 14, a result correction unit 15, a result notification unit 16, and a learning unit 17. Each part included in the abnormality determination device 1 may be configured by a computer provided with an arithmetic processing unit, a storage device, and the like, and the processing of each part may be executed by a program. This program is stored in a storage device included in the abnormality determination device 1, and can be recorded on a recording medium such as a magnetic disk, an optical disk, or a semiconductor memory, or can be provided through a network.

The signal processing unit 11 acquires the sensor data output from the sensors 31A to 31D, and processes the sensor data so that the abnormality detection unit 13 has a usable value. For example, the signal processing unit 11 detects when the missing value of the sensor data is complemented, the abnormal value of the sensor data is removed, the sensor data is corrected, the frequency characteristic analysis is performed by FFT or the like, and the reciprocating motion of the robot arm is executed. Converts outbound and inbound sensor data into feature data. The processing performed by the signal processing unit 11 is not limited to this. The signal processing unit 11 records the sensor data before processing, the sensor data after processing, various data obtained from the sensor data, and the data acquisition date and time in the storage unit 12.

The abnormality detection unit 13 detects an abnormality in each of the robots 3A to 3D based on the data recorded in the storage unit 12. For example, the abnormality detection unit 13 obtains the degree of abnormality from the data recorded in the storage unit 12, and determines that the abnormality is abnormal when the degree of abnormality is equal to or greater than a predetermined threshold value.

The correlation detection unit 14 detects whether or not there is a correlation between the abnormality detection state of the robot 3A for which the abnormality detection unit 13 has detected the abnormality and the abnormality detection state of the other robots 3B to 3D. The correlation of the abnormality detection state between the robots 3A and 3D is, for example, the correlation of the sensor data used for detecting the abnormality, the correlation of the degree of abnormality used for detecting the abnormality, the correlation of the timing when the abnormality is detected, or the correlation of the abnormality detection result. It is obtained by correlation. The correlation detected by the correlation detection unit 14 is not limited to this. The correlation detection unit 14 may detect the presence or absence of correlation in the abnormality detection state based on the learning result of the learning unit 17 described later.

The result correction unit 15 maintains or corrects the abnormality detection results of the robots 3A to 3D based on the correlation detection result of the correlation detection unit 14 and a predetermined rule. Specifically, the result correction unit 15 determines the abnormality detection result of the robot 3A when there is a correlation between the abnormality detection state of the robot 3A in which the abnormality is detected and the abnormality detection state of the other robots 3B to 3D. Correct from anomaly judgment to normal judgment.

If the result notification unit 16 determines that the abnormality is abnormal by the abnormality detection unit 13 and the result correction unit 15 does not correct the result to the normal determination, the result is reported to related persons such as workers, maintenance personnel, equipment managers, and supervisors. Notify the alarm. For example, the result notification unit 16 notifies an alarm using a display device such as a rotating light or a liquid crystal display, an audio device such as a buzzer or a speaker that emits a sound, and an information medium such as a telephone, an e-mail, or an SNS capable of transmitting a message. .. Even in the normal state when it is not determined to be abnormal, it may be notified that it is normal rather than abnormal, and the notification is not limited to only the abnormality. Specifically, it constantly notifies that it is in a normal state, and when an abnormality occurs, the content of the notification is switched from normal to abnormal.

The learning unit 17 is the result of the result correction unit 15 processing the abnormality detection results of the robots 3A to 3D in order to improve the accuracy and processing speed when the correlation detection unit 14 detects the presence or absence of the correlation in the abnormality detection state. To record and learn. The learning unit 17 may learn together with the data recorded in the storage unit 12.

Although the work robot has been described as an example of the abnormality determination target, the present invention is not limited to the work robot. For example, it can be applied to an automobile engine instead of a motor and a transmission instead of a reduction gear. In addition to this, it can be applied to moving objects in amusement parks and machine tools such as three-dimensional printers. It can be applied to any one having a rotation mechanism and a mechanism for transmitting it. It is an invention that can be used for all.

The abnormality determination method by the abnormality determination device 1 of the present embodiment will be described with reference to FIG. The processing shown in the flowchart of FIG. 2 is performed for each of the robots 3A to 3D. Hereinafter, an example in which the robot 3A is targeted for abnormality determination will be described.

In step S11, the signal processing unit 11 receives sensor data from the sensor 31A for detecting the state of the robot 3A to be determined as an abnormality, complements the deficiency in the sensor data, and mixes an abnormal value. Performs data preprocessing such as removal and correction. The processed data is stored in the storage unit 12.

In step S12, the abnormality detection unit 13 determines whether or not an abnormality / failure is predicted to occur in the robot 3A. For example, the abnormality detection unit 13 detects an abnormality using an unsupervised model that predicts an abnormality from a normal state based on sensor data received in the past, or obtains it from the current sensor data and obtains the degree of abnormality and the physical. By comparing with the threshold value of the degree of abnormality based on the phenomenon, it is determined whether or not the occurrence of an abnormal failure is predicted.

When the abnormality detection unit 13 does not detect an abnormality, the abnormality determination device 1 returns the process to step S11 and continues to acquire the sensor data.

When the abnormality detection unit 13 detects an abnormality, in step S13, the correlation detection unit 14 determines whether or not the abnormality detected by the robot 3A correlates with the other robots 3B to 3D. For example, the correlation detection unit 14 obtains the correlation between the degree of abnormality that is the source of the abnormality detection of the robot 3A and the degree of abnormality of the robot 3B.

FIG. 3 shows an example in which there is no correlation between the abnormalities of the two robots A and B, and FIG. 4 shows an example in which there is a correlation. In the examples of FIGS. 3 and 4, the horizontal axis represents time and the vertical axis represents the degree of abnormality, and the changes in the degree of abnormality of the robots A and B are plotted. The positions indicated by the arrows in FIGS. 3 and 4 are the positions where the robot A has detected an abnormality. In FIG. 3, when the degree of abnormality changes significantly in robot A, the degree of abnormality does not change in robot B. The correlation detection unit 14 determines that there is no correlation between the robots A and B in the abnormality detection state. In FIG. 4, when the degree of abnormality changes significantly in the robot A, the degree of abnormality also changes significantly in the robot B. Further, in FIG. 4, the change in the degree of abnormality of the robot A and the change in the degree of abnormality of the robot B are similar. The correlation detection unit 14 determines that there is a correlation between the robots A and B in the abnormality detection state. The determination of the correlation is not limited to the case where the occurrence times completely match, for example, and a time lag of several seconds to several minutes may be allowed in consideration of an error such as a timer shift.

If it is determined in step S13 that there is a correlation, in step S14, the result correction unit 15 corrects the abnormality determination for the robot 3A to a normal determination. Here, the probability of occurrence of an abnormal failure of the robot 3A is set to 0.1%. The probability of abnormal failure of robot 3B of the same type as robot 3A is also considered to be 0.1%. At this time, the probability that an abnormal failure will occur in the robot 3A and the robot 3B at the same time is calculated to be 0.0001%. In this way, it is extremely unlikely that an abnormal failure will occur in the robot 3A and the robot 3B at the same time. Therefore, the abnormal failure does not occur in each of the robot 3A and the robot 3B, but in the environment surrounding the robot 3A and the robot 3B. I think there is a cause. Therefore, the result correction unit 15 normally corrects the abnormality detection result of the robot 3A from the abnormality.

Subsequently, in step S15, the learning unit 17 records the result including the derivation contents of the correlation detection unit 14 and the result correction unit 15 on a recording medium such as a cloud or a server. The learning unit 17 may create a model used for learning the recorded results and obtaining the correlation of the degree of abnormality in order to improve the accuracy and the processing speed.

On the other hand, if it is determined in step S13 that there is no correlation, the result correction unit 15 does not correct the abnormality detection result. In step S16, the result notification unit 16 notifies the abnormality detection result as an alarm. When the result correction unit 15 corrects the abnormality detection result to make a normal determination, the result notification unit 16 may convey the corrected abnormality detection result as a status display instead of an alarm. Specifically, the result notification unit 16 displays the corrected abnormality detection result on a display device such as a liquid crystal display or an electric bulletin board.

The abnormality determination device 1 returns to step S11 and continues the abnormality determination process unless an operation for ending the abnormality determination process is input.

As described above, according to the abnormality determination device 1 of the present embodiment, the following effects can be obtained.

The abnormality detection unit 13 detects an abnormality in the robots 3A to 3D based on the sensor data acquired from the sensors 31A to 31D. The correlation detection unit 14 determines whether or not there is a correlation between the abnormality determination state of the robot 3A determined to be abnormal and the abnormality determination state of the robots 3B to 3D other than the robot 3A. When there is a correlation between the abnormality determination states, the result correction unit 15 corrects the abnormality detection result of the robot 3A from the abnormality determination to the normal determination. As a result, the abnormality determination device 1 can suppress erroneous detection of the abnormality of the robots 3A to 3D itself detected by the robots 3A to 3D.

[Second Embodiment]
The abnormality determination device of the second embodiment will be described with reference to FIG. In the second embodiment, the target for comparing the presence / absence of correlation of the abnormality detection state is limited to robots in the same process or similar work processes, and if there is a correlation, the abnormality detection result is corrected.

The abnormality determination device 1 of the second embodiment is different from the abnormality determination device 1 of the first embodiment in that it includes an extraction unit 18. Other configurations are the same as in the first embodiment. In the second embodiment, the extraction unit 18 extracts the sensor data and the degree of abnormality of the robot 3B in the same process or a similar work process as the robot 3A in which the abnormality is detected by the abnormality detection unit 13 from the storage unit 12. The correlation detection unit 14 detects the presence or absence of a correlation in the abnormality detection state between the robots 3A and 3B.

For example, as shown in FIG. 6, it is assumed that a plurality of robots A to D are arranged on the left and right sides of the production line. When the work is a symmetrical product such as an automobile, the robots A and B facing each other across the work perform the same work at the same time, and thus are robots in the same process. Similarly, the robots C and D are robots in the same process.

The same process is not limited to those that perform exactly the same work at the same time. Robots with similar work processes such as time, space, and work content may be robots of the same process. In the example of FIG. 6, since the robots A to D are operating at the same time zone and are working on the same production line in the same factory, the robots A to D are robots that perform similar work processes. Is. The presence or absence of correlation between the abnormality detection states may be compared with any combination of robots A to D. When there are multiple robots with similar work processes, the robot with the highest degree of similarity in time, space, and work content may be the target for comparing the presence or absence of correlation between the abnormality detection states.

The abnormality determination method by the abnormality determination device 1 of the present embodiment will be described with reference to FIG. 7. The flowchart of FIG. 7 is different from the first embodiment in that steps S23 and S24 are performed instead of step S13 of the flowchart of FIG.

In step S21, the signal processing unit 11 receives the sensor data from the sensor 31A and performs data preprocessing.

In step S22, the abnormality detection unit 13 determines whether or not an abnormality / failure is predicted to occur in the robot 3A.

When the abnormality detection unit 13 does not detect an abnormality, the abnormality determination device 1 returns the process to step S21 and continues to acquire the sensor data.

When the abnormality detection unit 13 detects an abnormality, in step S23, the extraction unit 18 determines whether or not other robots 3B to 3D in the same process as the robot 3A exist.

If there are no robots in the same process, the result correction unit 15 does not correct the abnormality detection result. In step S27, the result notification unit 16 notifies the abnormality detection result as an alarm.

When there are robots in the same process, the extraction unit 18 extracts the sensor data and the degree of abnormality of the robot 3B in the same process from the storage unit 12 and outputs them to the correlation detection unit 14. In step S24, the correlation detection unit 14 determines whether or not the abnormality detected by the robot 3A correlates with the robot 3B in the same process. When there are two or more robots 3B to 3D in the same process, the extraction unit 18 selects the robot 3B that performs the work process most similar to the work process of the robot 3A among the robots 3B to 3D in the same process. For example, the extraction unit 18 calculates the similarity between the work processes of the robots 3B to 3D and the work process of the robot 3A based on the time, space, and work content, and determines the robot 3B having the highest calculated similarity. select.

If it is determined in step S24 that there is a correlation, in step S25, the result correction unit 15 corrects the detected abnormality determination to a normal determination.

In step S26, the learning unit 17 records the result including the derived contents of the correlation detection unit 14 and the result correction unit 15 on a recording medium such as a cloud or a server.

On the other hand, if it is determined in step S24 that there is no correlation, the result correction unit 15 does not correct the abnormality detection result. In step S27, the result notification unit 16 notifies the abnormality detection result as an alarm.

The abnormality determination device 1 returns to step S21 and continues the abnormality determination process unless an operation for ending the abnormality determination process is input.

As described above, according to the abnormality determination device 1 of the present embodiment, the following effects can be obtained.

The correlation detection unit 14 selects the robot 3B having a work process similar to the work process of the robot 3A in which the abnormality detection unit 13 has detected the abnormality, and whether or not there is a correlation between the abnormality detection state of the robot 3A and the abnormality detection state of the robot 3B. To judge. Since the correlation is obtained between robots 3A and 3B with similar work processes, improvement in accuracy can be expected. Further, since the target for which the correlation is to be obtained is limited to the robot 3B having a similar work process, improvement in processing efficiency and reduction in processing time can be expected.

[Third Embodiment]
Next, the third embodiment will be described. In the third embodiment, when there are a plurality of robots in the same process, the target for which the presence / absence of correlation of the abnormality detection state is compared is changed in order to correct the abnormality detection result. The configuration of the abnormality determination device 1 of the third embodiment is the same as that of the second embodiment shown in FIG.

For example, it is assumed that the plurality of robots A to D shown in FIG. 6 are robots in the same process. In the third embodiment, when an abnormality is detected in the robot A, first, it is determined whether or not there is a correlation between the robot A and the robot B. If there is no correlation between robots A and B, it is determined whether or not there is a correlation between robot A and robot C. If there is no correlation between robots A and C, it is determined whether or not there is a correlation between robot A and robot D. If the abnormality detection state of the robot A correlates with any of the abnormality detection states of the robots B to D, the abnormality detection result is corrected.

The abnormality determination method by the abnormality determination device 1 of the present embodiment will be described with reference to FIG. The flowchart of FIG. 8 is different from the second embodiment in that step S37 is performed after NO in step S24 of the flowchart of FIG. 7 to determine whether or not there is a correlation with all robots in the same step.

In step S31, the signal processing unit 11 receives the sensor data from the sensor 31A and performs data preprocessing.

In step S32, the abnormality detection unit 13 determines whether or not an abnormality / failure is predicted to occur in the robot 3A.

When the abnormality detection unit 13 does not detect an abnormality, the abnormality determination device 1 returns the process to step S31 and continues to acquire the sensor data.

When the abnormality detection unit 13 detects an abnormality, in step S33, the extraction unit 18 determines whether or not other robots 3B to 3D in the same process as the robot 3A exist.

If there are no robots in the same process, the result correction unit 15 does not correct the abnormality detection result. In step S38, the result notification unit 16 notifies the abnormality detection result as an alarm.

When there are robots in the same process, the extraction unit 18 extracts the sensor data and the degree of abnormality of the robot 3B in the same process from the storage unit 12 and outputs them to the correlation detection unit 14. In step S34, the correlation detection unit 14 determines whether or not the abnormality detected by the robot 3A correlates with the robot 3B in the same process.

If it is determined in step S34 that there is no correlation, in step S37, the correlation detection unit 14 determines whether or not the correlation of the abnormality detection state has been obtained for all the robots 3B to 3D in the same step.

When there are robots 3B to 3D in the same process for which the correlation of the abnormality detection state is not obtained, the correlation detection unit 14 selects another robot 3B to 3D in the same process and performs the process in step S34.

When the correlation of the abnormality detection state is obtained for all the robots 3B to 3D in the same process, the result correction unit 15 does not correct the abnormality detection result. In step S38, the result notification unit 16 notifies the abnormality detection result as an alarm.

On the other hand, when it is determined in step S34 that there is a correlation, the result correction unit 15 corrects the detected abnormality determination to a normal determination in step S35.

Subsequently, in step S36, the learning unit 17 records the result including the derived contents of the correlation detection unit 14 and the result correction unit 15 on a recording medium such as a cloud or a server.

The abnormality determination device 1 returns to step S31 and continues the abnormality determination process unless an operation for ending the abnormality determination process is input.

As described above, according to the abnormality determination device 1 of the present embodiment, the following effects can be obtained.

The correlation detection unit 14 selects the robot 3B having a work process similar to the work process of the robot 3A in which the abnormality detection unit 13 has detected the abnormality, and whether or not there is a correlation between the abnormality detection state of the robot 3A and the abnormality detection state of the robot 3B. To judge. When there is a correlation with the abnormality determination state of the selected robot 3B, the result correction unit 15 corrects the abnormality detection result of the robot 3A from the abnormality determination to the normal determination. If there is no correlation, the correlation detection unit 14 selects another robot 3C, 3D having a similar work process. Since the correlation detection unit 14 obtains the correlation with a plurality of robots 3B to 3D, improvement in accuracy can be expected.

1 ... Abnormality determination device 11 ... Signal processing unit 12 ... Storage unit 13 ... Abnormality detection unit 14 ... Correlation detection unit 15 ... Result correction unit 16 ... Result notification unit 17 ... Learning unit 18 ... Extraction unit 3A to 3D ... Robot 31A to 31D … Sensor

Claims (6)

An abnormality detection unit that detects an abnormality in each of the plurality of work devices based on sensor data acquired from each of the plurality of work devices.
When the abnormality detection unit detects an abnormality in any of the plurality of work devices, the abnormality detection state of the first work device in which the abnormality is detected and the abnormality of the second work device other than the first work device are abnormal. An abnormality determination device including a result correction unit that corrects the detection result of the first work device based on the detection state. Claim 1 is characterized in that the result correction unit corrects the detection result when there is a correlation between the abnormality detection state of the first work device and the abnormality detection state of the second work device. The abnormality determination device described in 1. The abnormality determination device according to claim 1 or 2, wherein the work process of the first work equipment and the work process of the second work equipment are similar to each other. Claim 1 is characterized in that, among the plurality of work devices, the result correction unit uses the work device that performs a work process most similar to the work process of the first work device as the second work device. The abnormality determination device according to any one of 3 to 3. When the result correction unit does not correct the detection result based on the abnormality detection state of the first work device and the abnormality detection state of the second work device, the result correction unit is different from the plurality of work devices. The abnormality determination device according to any one of claims 1 to 4, wherein the detection result is corrected by selecting. Based on the sensor data acquired from each of the plurality of work devices, each abnormality of the plurality of work devices is detected, and the abnormality is detected.
When any abnormality of the plurality of work devices is detected, the abnormality is detected based on the abnormality detection state of the first work device and the abnormality detection state of the second work device other than the first work device. An abnormality determination method, characterized in that the detection result of the first work device is corrected.

Images