Machine learning apparatus, machine learning method, and industrial machine

Patent metadata

Field	Value
Publication number	US-11748659-B2
Application number	US-202017027480-A
Country	US
Kind code	B2
Filing date	Sep 21, 2020
Priority date	Sep 27, 2019
Publication date	Sep 5, 2023
Grant date	Sep 5, 2023

Abstract

A machine learning apparatus determines a control parameter of an active vibration isolation apparatus on which an industrial machine is mounted. The industrial machine includes a movable part, a drive source that drives the movable part, and a drive source control section that controls the drive source to position the movable part at a command position. The machine learning apparatus includes: an acquiring section that acquires, as teacher data, a positional deviation, which is a difference between the command position and an actual position of the movable part; a storage section that stores a learning model that outputs the control parameter corresponding to a state quantity concerning the industrial machine; and a learning section that updates the learning model using the teacher data.

First claim

What is claimed is:

1. A machine learning apparatus that determines a control parameter of an active vibration isolation apparatus on which an industrial machine is mounted, wherein
the industrial machine includes a movable part, a drive source configured to drive the movable part, and a drive source control section configured to control the drive source to position the movable part at a command position,
the machine learning apparatus comprising:
an acquiring section configured to acquire, as teacher data, a positional deviation, which is a difference between the command position and an actual position of the movable part;
a storage section configured to store a learning model configured to output the control parameter corresponding to a state quantity concerning the industrial machine; and
a learning section configured to update the learning model using the teacher data.

2. The machine learning apparatus according to claim 1 , further comprising:
a control parameter determining section configured to determine the control parameter corresponding to the state quantity, using the learning model updated by the learning section; and
an output section configured to output, to the active vibration isolation apparatus, the control parameter determined by the control parameter determining section.

3. The machine learning apparatus according to claim 1 , wherein
the acquiring section is configured to further acquire, as the teacher data, information indicating vibration of the active vibration isolation apparatus.

4. The machine learning apparatus according to claim 1 , wherein
the control parameter includes at least any one of an internal pressure of an air spring configured to lift up a vibration isolation platform provided in the active vibration isolation apparatus, a drive amount of a servo valve configured to actively control the air spring, a target position of an air spring unit that includes the air spring, control feedback gain of the active vibration isolation apparatus, and a setting value for a filter provided in the active vibration isolation apparatus.

5. The machine learning apparatus according to claim 1 , wherein
the state quantity includes at least any one of strength of a floor on which the industrial machine is installed via the active vibration isolation apparatus, a positional relationship between the industrial machine and another machine, mass of a workpiece, an ambient temperature of the industrial machine, wind pressure experienced by the industrial machine, sound surrounding the industrial machine, and vibration applied to the industrial machine.

6. The machine learning apparatus according to claim 1 , wherein
the drive source is a servo motor.

7. The machine learning apparatus according to claim 1 , wherein
the industrial machine is an ultra-high precision machine tool configured to machine a workpiece with a machine accuracy of 100 nm or less according to a machining command.

8. An industrial machine comprising the machine learning apparatus according to claim 1 .

9. A machine learning method for determining a control parameter of an active vibration isolation apparatus on which an industrial machine is mounted, wherein
the industrial machine includes a movable part, a drive source configured to drive the movable part, and a drive source control section configured to control the drive source to position the movable part at a command position,
the machine learning method comprising:
an acquiring step of acquiring, as teacher data, a positional deviation, which is a difference between the command position and an actual position of the movable part; and
a step of updating a learning model using the teacher data, the learning model being configured to output the control parameter corresponding to a state quantity concerning the industrial machine.

10. The machine learning method according to claim 9 , further comprising:
a step of determining the control parameter corresponding to the state quantity, using the learning model updated in the step of updating the learning model; and
a step of outputting, to the active vibration isolation apparatus, the control parameter determined in the step of determining the control parameter.

11. The machine learning method according to claim 9 , wherein
in the acquiring step, information indicating vibration of the active vibration isolation apparatus is further acquired as the teacher data.

12. The machine learning method according to claim 9 , wherein
the control parameter includes at least any one of an internal pressure of an air spring configured to lift up a vibration isolation platform provided in the active vibration isolation apparatus, a drive amount of a servo valve configured to actively control the air spring, a target position of an air spring unit that includes the air spring, control feedback gain of the active vibration isolation apparatus, and a setting value for a filter provided in the active vibration isolation apparatus.

13. The machine learning method according to claim 9 , wherein
the state quantity includes at least any one of strength of a floor on which the industrial machine is installed via the active vibration isolation apparatus, a positional relationship between the industrial machine and another machine, mass of a workpiece, an ambient temperature of the industrial machine, wind pressure experienced by the industrial machine, sound surrounding the industrial machine, and vibration applied to the industrial machine.

Assignees

• Fanuc Corp

Inventors

• Li Zhenxing
• Minami Hiroshi
• Hada Keita
• Maeda Kazuomi

Classifications

• G06N20/00 Primary

  Machine learning · CPC title

• G05B19/404

  characterised by control arrangements for compensation, e.g. for backlash, overshoot, tool offset, tool wear, temperature, machine construction errors, load, inertia (G05B19/19, G05B19/41 take precedence) · CPC title

• G05B2219/37634

  By measuring vibration · CPC title

• G05B2219/49048

  Control of damping of vibration of machine base · CPC title

• G05B2219/49065

  Execute learning mode first for determining adaptive control parameters · CPC title