ニューラルネットワークによる超微小押込み曲線を用いた多層薄膜材の機械的特性評価

概要

論文の詳細を見る
This paper describes the application of neural network for nano-indentation testing to evaluate mechanical properties of multi-layered thin films. The properties of films are evaluated from a set of load-depth curves which is obtained by indentation using spherical indenters having different radii. In this study, finite element simulation of indentation is conducted to provide learning data of the network. Two multi-layered networks are independently used to identify the properties of films. By using the first network, the ranges of estimation are roughly estimated, and the appropriate properties can be determined by the second network. It is demonstrated that Young's moduli of 4-layered films can be successfully determined by the proposed method. It is also shown that both of Young's modulus and thickness can be identified 2-layered films.
社団法人日本機械学会の論文
2002-02-25