낙하 충격에 의한 탄소-케블라 하이브리드 직물의 전압 강하 특성과 머신 러닝을 통한 충격체 형상 예측

초록

Conductivity of carbon fibers can be utilized in a variety of applications such as the damage detection in special suits and composites requiring structural health monitoring. In this study, a methodology capable of predicting the impactor shapes was proposed by applying the voltage drop information induced by the damage of carbon-Kevlar hybrid fabrics to a decision tree-based random forest algorithm. drop impact tests were performed on the carbon-Kevlar hybrid fabric specimens according to the impact shapes and incident angles. Using the Gini index of the classification and regression tree (CART) statistical technique, important variables of the impact shape prediction criteria were analyzed. The validity of the technique was verified by out of bag (OOB) error estimation and three-fold cross validation. The shape of the impactor was precisely predicted by the unknown impactor from the voltage drop data, which are not included in the training process of the random forest. This study is significant in that it predicts the shape of the initial impactor through the machine learning technique by reflecting a multitude of object signals rather than based on specific parameters.

키워드