Machine learning system and magnetizer for motor

Patent metadata

Field	Value
Publication number	US-10061276-B2
Application number	US-201615273704-A
Country	US
Kind code	B2
Filing date	Sep 23, 2016
Priority date	Sep 30, 2015
Publication date	Aug 28, 2018
Grant date	Aug 28, 2018

Abstract

A machine learning system according to an embodiment of the present invention includes a state observer for observing the winding temperature, winding resistance, current value, and rotor magnetic flux density of a magnetization unit having a magnetizing yoke and windings; a reward calculator for calculating a reward from the rotor magnetic flux density obtained by the state observer; and a learning unit for updating an action value table based on a magnetization rate calculated from the rotor magnetic flux density and a target magnetic flux density, the winding temperature, and the winding resistance.

First claim

What is claimed is:

1. A magnetizer for manufacturing rotors for motors, the magnetizer comprising:
a magnetization unit having a magnetizing yoke and windings configured to magnetize the rotors; and
a machine learning system, comprising:
a state observer configured to observe a winding temperature, a winding resistance, a current value, and a rotor magnetic flux density of the magnetization unit having the magnetizing yoke and windings; and
a processor configured to
calculate a reward from the rotor magnetic flux density obtained by the state observer, and
update an action value table based on a magnetization rate calculated from the rotor magnetic flux density and a target magnetic flux density, the winding temperature, and the winding resistance,
wherein the processor is configured to perform an arithmetic operation of a state variable observed by the state observer in a multilayer structure, and update the action value table in real time.

2. The magnetizer according to claim 1 , wherein the processor is further configured to determine a voltage command based on the action value table.

3. The magnetizer according to claim 1 , wherein the processor is further configured to update the action value table using an action value table updated by another machine learning system.

Assignees

• Fanuc Corp

Inventors

• Funakubo Makoto

Classifications

• H02P21/06

  Rotor flux based control involving the use of rotor position or rotor speed sensors · CPC title

• G05B13/0265 Primary

  the criterion being a learning criterion · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title

• H01F13/003

  Methods and devices for magnetising permanent magnets (permanent magnets H01F7/02) · CPC title