Machine learning apparatus and coil producing apparatus

Patent metadata

Field	Value
Publication number	US-10019674-B2
Application number	US-201615280805-A
Country	US
Kind code	B2
Filing date	Sep 29, 2016
Priority date	Sep 30, 2015
Publication date	Jul 10, 2018
Grant date	Jul 10, 2018

Abstract

A machine learning apparatus includes a state observing unit for observing a state variable comprised of at least one of an actual dimension value, a resistance actual value, etc., and at least one of a dimension command value, a resistance command value, etc., and an execution time command value for a program, and a learning unit for performing a learning operation by linking at least one of an actual dimension value, a resistance actual value, etc., to at least one of a dimension command value, a resistance command value, etc., observed by the state observing unit, and an execution time command value for the program.

First claim

What is claimed is:

1. A control apparatus for controlling a winding machine for forming a coil, the control apparatus a machine learning apparatus communicable with the winding machine and configured to learn an operation for forming the coil by the winding machine, the machine learning apparatus comprising:
a state observing unit for observing a state variable comprised of at least one of an actual dimension value, a resistance actual value, and a wire rod used amount of the coil formed by the winding machine, and a program execution time actual value, and at least one of a dimension command value, a resistance command value, a turn number command value, a winding speed command value, and a tension command value for the coil, which are instructed by a program for the winding machine, and an execution time command value for the program; and
a learning unit for learning by linking at least one of the actual dimension value, the resistance actual value, and the wire rod used amount of the coil observed by the state observing unit, and the program execution time actual value to at least one of the dimension command value, the resistance command value, the turn number command value, the winding speed command value, and the tension command value for the coil observed by the state observing unit, and the execution time command value for the program, wherein the learning unit comprises:
a reward computing unit for computing a reward based on at least one of the actual dimension value, the resistance actual value, and the wire rod used amount of the coil observed by the state observing unit, and the program execution time actual value; and
a function updating unit for updating a function for deciding, from the state variable at present, based on the reward computed by the reward computing unit, at least one of the dimension command value, the resistance command value, the turn number command value, the winding speed command value, and the tension command value for the coil, and the execution time command value for the program,
wherein the learning unit is configured to compute the state variable observed by the state observing unit in a multilayer structure, to update the function on a real-time basis, and

a decision-making unit for deciding, in accordance with the updated function, an optimal value of at least one of the dimension command value, the resistance command value, the turn number command value, the winding speed command value, and the tension command value for the coil, and the execution time command value for the program,
wherein the control apparatus is configured to control the winding machine in accordance with the program having the optimal value decided by the decision-making unit.

2. The control apparatus according to claim 1 , wherein the reward computing unit is configured to
increase the reward when the actual dimension value, the resistance actual value, and the wire rod used amount of the coil, and the program execution time actual value remain within respective allowable ranges, and
decrease the reward when the actual dimension value, the resistance actual value, and the wire rod used amount of the coil, and the program execution time actual value are outside of the respective allowable ranges.

3. The control apparatus according to claim 1 , wherein the function updating unit is configured to update the function by using a function updated by a function updating unit of a different machine learning apparatus.

4. A coil producing apparatus, comprising:
a winding machine for forming a coil; and
a control apparatus according to claim 1 for controlling the winding machine.

Assignees

• Fanuc Corp

Inventors

• Sugimoto Yasunori

Classifications

• G06N3/042

  Knowledge-based neural networks; Logical representations of neural networks · CPC title

• H01F41/00

  Apparatus or processes specially adapted for manufacturing or assembling magnets, inductances or transformers; Apparatus or processes specially adapted for manufacturing materials characterised by their magnetic properties · CPC title

• G06N99/005 Primary

  Physics · mapped topic

• G06N5/045

  Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence · CPC title

• G06N3/0427

  Physics · mapped topic