Machine learning device and machine learning method for learning fault prediction of main shaft or motor which drives main shaft, and fault prediction device and fault prediction system including machine learning device

Patent metadata

Field	Value
Publication number	US-11521105-B2
Application number	US-201715481575-A
Country	US
Kind code	B2
Filing date	Apr 7, 2017
Priority date	Apr 8, 2016
Publication date	Dec 6, 2022
Grant date	Dec 6, 2022

Abstract

A machine learning device which learns fault prediction of one of a main shaft of a machine tool and a motor driving the main shaft, including a state observation unit observing a state variable including at least one of data output from a motor controller controlling the motor, data output from a detector detecting a state of the motor, and data output from a measuring device measuring a state of the one of the main shaft and the motor; a determination data obtaining unit obtaining determination data upon determining one of whether a fault has occurred in the one of the main shaft and the motor and a degree of fault; and a learning unit learning the fault prediction of the one of the main shaft and the motor in accordance with a data set generated based on a combination of the state variable and the determination data.

First claim

What is claimed is:

1. A machine learning device for learning fault prediction of one of a main shaft of a machine tool and a motor which drives the main shaft, the machine learning device comprising:
a processor configured to
observe a state variable comprising at least one of data output from a motor controller configured to control the motor, data output from a detector configured to detect a state of the motor, and data including a state of the one of the main shaft and the motor,
obtain determination data upon determining one of whether a fault has occurred in the one of the main shaft and the motor and a degree of fault, and
learn the fault prediction of the one of the main shaft and the motor in accordance with a data set generated based on a combination of the state variable and the determination data, wherein

the processor is configured to, in response to obtaining the determination data indicating a fault of the one of the main shaft and the motor, update a conditional equation used for the fault prediction by weighting the determination data in the data set in correlation with a length of time from when the determination data is obtained until the fault occurs,
the processor is configured to
perform a health check operation causing the main shaft to perform a predetermined operation pattern during a predetermined time interval, and
observe the state variable in at least one of an accelerated state of the main shaft, a constant velocity-operated state of the main shaft, and one of a decelerated state of the main shaft and a coasted state of the main shaft upon cutoff of a driving force, in the predetermined operation pattern, to learn the fault prediction of the one of the main shaft and the motor, and

the health check operation includes causing the main shaft to be in the coasted state to observe the state variable and identify a symptom of the fault different from that in the decelerated state.

2. The machine learning device according to claim 1 , wherein
the data output from the motor controller comprises a torque command value and a velocity command value for driving the motor,
the detector comprises at least one of a current detector configured to detect a current of the motor and a velocity detector configured to detect a velocity of the motor, and
the processor is configured to observe the data output from at least one of
a vibration sensor configured to measure vibration of the one of the main shaft and the motor,
a sound collection microphone configured to measure a sound in vicinity of the one of the main shaft and the motor, and
a temperature sensor configured to measure a temperature in the vicinity of the one of the main shaft and the motor.

3. The machine learning device according to claim 1 , wherein the processor is configured to learn the fault prediction of the one of the main shaft and the motor in accordance with a plurality of data sets generated for the one of the main shaft and the motor.

4. The machine learning device according to claim 1 , wherein the processor is configured to learn a normal state only during a certain period and detect occurrence of a fault.

5. The machine learning device according to claim 1 , wherein the processor is configured to obtain a current state variable of the state variable via a network.

6. A fault prediction device for predicting a fault of the one of the main shaft and the motor, the fault prediction device comprising the machine learning device according to claim 1 and configured to output fault information indicating one of whether a fault has occurred in the one of the main shaft and the motor and a degree of fault, in response to input of a current state variable of the state variable, based on a learning result obtained by the machine learning device in accordance with the data set.

7. The fault prediction device according to claim 6 , wherein the machine learning device is configured to learn the fault prediction of the one of the main shaft and the motor again in accordance with an additional data set generated based on a combination of the current state variable and the determination data.

8. The fault prediction device according to claim 6 , wherein the machine learning device is located on a cloud server.

9. The fault prediction device according to claim 6 , further comprising a controller including the machine learning device and configured to control the machine tool.

10. The fault prediction device according to claim 6 , wherein the machine tool comprises a plurality of machine tools, and the learning result obtained by the machine learning device is shared by the plurality of machine tools.

11. A fault prediction system comprising:
the fault prediction device according to claim 8 ;
the detector;
a sensor or a microphone configured to measure the state of the one of the main shaft and the motor; and
a display or an alarm configured to notify an operator of the fault information.

12. The fault prediction system according to claim 11 , wherein a time at which the display or the alarm is configured to notify the operator of the fault information satisfies at least one of precedence to a time defined by a first predetermined period preceding a time at which a fault occurs, and subsequence to a time defined by a second predetermined period preceding the time at which a fault occurs.

13. A machine learning method of learning fault prediction of one of a main shaft of a machine tool and a motor which drives the main shaft, the method comprising:
observing a state variable comprising at least one of data output from a motor controller which controls the motor, data output from a detector which detects a state of the motor, and data including a state of the one of the main shaft and the motor;
obtaining determination data upon determining one of whether a fault has occurred in the one of the main shaft and the motor and a degree of fault;
learning the fault prediction of the one of the main shaft and the motor in accordance with a data set generated based on a combination of the state variable and the determination data, and
in response to obtaining the determination data indicating a fault of the one of the main shaft and the motor, updating a conditional equation used for the fault prediction by weighting the determination data in the data set in correlation with a length of time from when the determination data is obtained until the fault occurs, wherein
the method further comprises
performing a health check operation causing the main shaft to perform a predetermined operation pattern during a predetermined time interval, and
observing the state variable in at least one of an accelerated state of the main shaft, a constant velocity-operated state of the main shaft, and one of a decelerated state of the main shaft and a coasted state of the main shaft upon cutoff of a driving force, in the predetermined operation pattern, to learn the fault prediction of the one of the main shaft and the motor, and

the health check operation includes causing the main shaft to be in the coasted state to observe the state variable and identify a symptom of the fault different from that in the decelerated state.

Assignees

• Fanuc Corp

Inventors

• Kamiya Yohei
• Yamada Yasuo

Classifications

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G01M15/14

  Testing gas-turbine engines or jet-propulsion engines · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G01N29/4481

  Neural networks · CPC title

• G06N5/04

  Inference or reasoning models · CPC title