Title:
MACHINE LEARNING METHOD, MACHINE LEARNING DEVICE, MACHINE LEARNING PROGRAM, COMMUNICATION METHOD AND CONTROL DEVICE
Document Type and Number:
Japanese Patent JP2023062867
Kind Code:
A
Abstract:
To easily determine a proper CIP processing condition for a processing object.SOLUTION: A machine learning method calculates a reward for a determination result of an isotropic pressurizing processing condition on the basis of a state variable including at least one physical amount about a processing object and at least one isotropic pressurizing processing condition, updates a function for determining at least one isotropic pressurizing processing condition from the state variable on the basis of the reward, and determines the isotropic pressurizing processing condition in which the most reward can be obtained by repeating the update of the function. The isotropic pressurizing processing condition is at least one of a first parameter about the processing object, a second parameter about a preceding step of the isotropic pressurizing processing and a third parameter about an operation condition of an isotropic pressurizing device. At least one physical amount is at least one of physical amounts about densification and pressure-powderization of the processing object.SELECTED DRAWING: Figure 1
Inventors:
ITO HIROYUKI
MINAMINO TOMOYA
SHIRAKASHI HIROSHI
KISHI CHIKAKAZU
MIZOGAMI TADATAKA
MIYASHITA YASUHIDE
MINAMINO TOMOYA
SHIRAKASHI HIROSHI
KISHI CHIKAKAZU
MIZOGAMI TADATAKA
MIYASHITA YASUHIDE
Application Number:
JP2021173018A
Publication Date:
May 09, 2023
Filing Date:
October 22, 2021
Export Citation:
Assignee:
KOBE STEEL LTD
International Classes:
B30B11/00; G06N3/02; G06N20/00
Attorney, Agent or Firm:
Masataka Kotani
Takashi Gyomu
Takashi Gyomu