Damage?sensitive feature extraction with stacked autoencoders for unsupervised damage detection

In most real-world monitoring scenarios, the lack of measurements from damaged conditions requires application unsupervised approaches, mainly ones based on modal features estimated raw vibration data through traditional system identification methods. Although numerous successful applications using parameters have been reported, they demonstrated to be insufficient estimate a robust set damage-sensitive features. Inspired by idea compressed sensing and deep learning, an intelligent two-level feature extraction approach stacked autoencoders over pre-processed is proposed. This procedure can improve performance damage detection classifiers compressing into smaller highly informative when considering information entropy metrics. The proposed technique demonstrates significant improvement in classification approaches evaluated well-known sets Z-24 Bridge, where several scenarios were carried out under rigorous operational environmental effects.