AE法による繊維強化金属(FRM)の破壊機構の解明と疲労損傷評価 : 第2報,ニューラルネットワークによる疲労損傷度の評価

概要

論文の詳細を見る
FRM is widely used as the structural member. However, a fatigue which occupies much of the fracture accidents of the structures cannot be avoided and the growth mechanism of the fatigue damage is very complicated. Our aim is to establish the evaluation method for the damage and the fracture mechanism of the fatigue. In the previous paper, the method that evaluates the fracture types was presented. As the results, the following things became clear: (1) the feature spaces of the fracture types are represented by the correlation dimensions and the largest lyapunov exponents of the AE signals; (2) it became clear that the recognition of each fracture type is evaluated by the mahalanobis distance for each feature space. In this paper, the method that evaluates the fatigue damage by the neural network is presented. The network has been configured by learning the pattern information which are expressed by the relationship between the input (i.e. the generation degree of each fracture type and AE parameters) and output data (i.e. the fatigue damage). And the validity of the network is confirmed.
2008-11-25