1832 超音波波形の特徴を考慮したニューラルネットワークによる鋳鉄の摩擦圧接部の評価(G03-5 ニューラルネットワーク・非破壊検査,G03 材料力学)

概要

論文の詳細を見る
The report describes classification due to ultrasonic inspection of the frictional welding of cast iron using a neural network. Specimens with different frictional condition were prepared and echoes from the frictional welded location were processed in computer. The neural network learned wave form data and classified to five different conditions taking into account of the characteristic of the wave form. Data set of the neural network was obtained from weibull distribution of input data. Classification performance was evaluated using the data set. Higher classification was obtained by using weibull distribution in compared to other wave form learning.
社団法人日本機械学会の論文
2005-09-18