Deep-learning based damage sensing of carbon fiber/polypropylene composite via addressable conducting network

Abstract

In this work, damage sensing of carbon fiber reinforced polymer composite (CFRP) was conducted based on an addressable conducting network (ACN). To improve the accuracy of damage detection, a deep learning-based damage sensing system was developed. The data for deep learning were generated using a resist network model based on Kirchhoff's law. The generated data was verified through finite element analysis. Then, the Artificial Neural Network (ANN) deep learning algorithm was used for damage detection and evaluation. The accuracy of damage sensing was improved by applying the resist network model that considered not only delamination but also the damage of the carbon fiber. As a result, damage detection of CFRP was performed with a high accuracy rate of about 95%.