MACHINE LEARNING METHOD

• Main Authors: MAGAWA KIYOSHI, KAWAMURA RYOHEI, FUJIWARA YUTA, UENISHI KAZUKI, KAWAHARA TAKAHITO, ASAI MIKIO, TAKAHASHI YOSHITAKA
• Format: Patent
• Language: English; Japanese
• Published: 10.11.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIRCHEMICAL OR PHYSICAL PROPERTIES; MEASURING; PHYSICS; TESTING

To provide a machine learning method capable of reducing time for conducting an abnormality inspection of an evaluation object after a corrosion test.SOLUTION: A machine learning method includes: a dataset acquisition step of acquiring a dataset including a set of explanatory variables and objective variables, using a weight of an evaluation object before a corrosion test and quantization information of three-dimensional shape information showing a three-dimensional shape of a surface of the evaluation object after the corrosion test as the explanatory variables and a result of an abnormality inspection of the evaluation object after the corrosion test as the objective variables; and a learning step of conducting machine learning using the dataset to construct a learning model for conducting an abnormality inspection of an evaluation object after a corrosion test.SELECTED DRAWING: Figure 1 【課題】腐食試験後の評価対象物の異常点検を行う時間を短縮できる機械学習方法を提供する。【解決手段】本発明の機械学習方法は、腐食試験前の評価対象物の重量と、上記腐食試験後の上記評価対象物の表面の三次元形状を表す三次元形状情報の定量化情報とを説明変数とし、上記腐食試験後の上記評価対象物の異常点検の結果を目的変数として、上記説明変数及び上記目的変数の組からなるデータセットを取得するデータセット取得工程と、上記データセットを用いて機械学習を実施することにより、腐食試験後の評価対象物の異常点検を行うための学習モデルを構築する学習工程と、を備えることを特徴とする。【選択図】図１