Machine learning device, machined state prediction device, and control device

Patent metadata

Field	Value
Publication number	US-12326698-B2
Application number	US-202117792339-A
Country	US
Kind code	B2
Filing date	Jan 25, 2021
Priority date	Jan 31, 2020
Publication date	Jun 10, 2025
Grant date	Jun 10, 2025

Abstract

A machine learning device is provided with: an input data obtaining unit that, in burnishing process in which a processing surface of an arbitrary workpiece is surface-treated with an arbitrary tool, obtains as input data processing information including information of the workpiece prior to the burnishing process and information of a processing condition; a label obtaining unit that obtains label data indicating processed state information including a processed state of the workpiece after the burnishing process and surface roughness of the workpiece when the processed state is normal; and a learning unit that carries out supervised learning using the input data and the label data thus obtained to generate a learned model to which processing information of an upcoming burnishing process is input and which outputs processed state information for the burnishing process.

First claim

The invention claimed is:

1. A machine learning device, comprising:
a non-transitory memory configured to store a program;
a processor configured to execute the program stored on the memory to cause the machine learning device to:
obtain, as input data, machining information for burnishing in which surface treatment was performed by pressing an arbitrary tool against a machined surface of a workpiece, wherein the machining information including at least information on the workpiece before the burnishing and information on a machining condition for the burnishing;
obtain label data indicating machined state information including a machined state of the workpiece after the burnishing and a surface roughness of the workpiece in a case where the machined state is normal; and
use the obtained input data and the obtained label data to execute supervised learning and generate a learned model that takes as an input machining information regarding burnishing to be performed and outputs machined state information for the burnishing to be performed,
wherein the information on the workpiece before the burnishing includes surface roughness before machining of the workpiece,
the machined state information includes a surface roughness after machining in the case where the machined state is normal,
the machined state information outputted by the learned model is utilized to output an instruction to execute burnishing or an instruction to change the machining information, and
further the machined state information outputted by the learned model is utilized to generate an operational command which are utilized by a machine tool to execute burnishing on the machining target workpiece.

2. The machine learning device according to claim 1 , wherein
the information on the workpiece includes at least one of material, hardness, thickness, and surface roughness before machining of the workpiece, and
the information on the machining condition for the burnishing includes at least one of a relative rotation speed for between a tool and a workpiece, a relative feedrate for between the tool and the workpiece, and an amount of rolling compaction.

3. The machine learning device according to claim 1 , wherein the machined state information, for the workpiece after the burnishing, includes at least one of whether a machined state indicating no waviness or deformation is normal and a surface roughness after machining in the case where the machined state is normal.

4. A machined state prediction device, comprising:
a non-transitory memory configured to store a program;
a processor configured to execute the program stored on the memory to cause the machined state prediction device to:
perform the machined state prediction using a learned model that is generated by the machine learning device according to claim 1 and is configured to be inputted with machining information regarding burnishing to be performed and output the machined state information for the burnishing to be performed;
obtain from a machine tool, before burnishing, machining information that includes information on a machining condition for the burnishing to be performed and information on a machining target workpiece;
by inputting to the learned model the obtained machining information, predict the machined state information that is for the burnishing to be performed and is to be outputted by the learned model, and
in a case where the machined state will be normal, determine whether the surface roughness after machining is within the required accuracy.

5. The machined state prediction device according to claim 4 , wherein
the information on the machining target workpiece includes at least one of material, hardness, thickness, and surface roughness before machining of the workpiece, and
the information on the machining condition for the burnishing includes at least one of a relative rotation speed for between a tool and a workpiece, a relative feedrate for between the tool and the workpiece, and an amount of rolling compaction.

6. The machined state prediction device according to claim 4 , wherein the machined state information, for the machining target workpiece after the burnishing to be performed, includes at least one of whether a machined state indicating no waviness or deformation is normal and a surface roughness after machining in a case where the machined state is normal.

7. The machined state prediction device according to claim 6 , wherein the processor is further configured to execute the program to further cause the machined state prediction device to: compare the predicted surface roughness after machining with a pre-set required accuracy, and determine whether the surface roughness after machining is within the required accuracy.

8. The machined state prediction device according to claim 4 , wherein
the learned model is stored in a server that is connected so as to be accessible from the machined state prediction device via a network.

9. A machined state prediction device comprising:
the machine learning device according to claim 1 ;
a non-transitory memory configured to store a program;
a processor configured to execute the program stored on the memory to cause the machined state prediction device to:
perform the machined state prediction using the learned model that is generated by the machine learning device and is configured to be inputted with machining information regarding burnishing to be performed and output the machined state information for the burnishing to be performed;
obtain from a machine tool, before burnishing, machining information that includes information on a machining condition for the burnishing to be performed and information on a machining target workpiece;
by inputting to the learned model the obtained machining information, predict the machined state information that is for the burnishing to be performed and is to be outputted by the learned model, and
in a case where the machined state will be normal, determine whether the surface roughness after machining is within the required accuracy.

10. A control device comprising: the machined state prediction device according to claim 4 .

Assignees

• Fanuc Corp

Inventors

• Ogino Hideo

Classifications

• G05B13/026

  using a predictor · CPC title

• B24B51/00

  Arrangements for automatic control of a series of individual steps in grinding a workpiece (if applicable to other machine tools, G05B takes precedence) · CPC title

• B24B39/00

  Burnishing machines or devices, i.e. requiring pressure members for compacting the surface zone (modifying the physical properties or structure of metal by burnishing C21D7/08, C22F1/00); Accessories therefor · CPC title

• B24B49/12

  involving optical means · CPC title

• B24B39/04

  designed for working external surfaces of revolution · CPC title