Feature extraction based on dynamic response measurements for structural damage identification: a comparative study

Abstract

ABSTRACT Sudden unexpected loads, such as wind and earthquakes, can cause significant damage and notable changes in the characteristics of structures. Additionally, unfavorable environmental conditions can lead to erosion and deterioration of materials and structural integrity. Therefore, structural health monitoring (SHM) and damage identification have become critical concerns in civil engineering. Feature extraction from dynamic response measurements is a crucial aspect of structural damage identification. Various methods and techniques have been developed to extract features for identifying damage and monitoring the health of structures. This paper presents a comparative study of three different damage-sensitive features: resonant frequency, discrete wavelet transform component energy (DWTCE), and wavelet packet transform component energy (WPTCE), which are used in damage identification. The displacement and acceleration responses were used in the feature extraction process. This study assesses the sensitivity of each feature to structural damage using a derivative-based method. The results showed that WPTCE extracted from the acceleration response data can effectively capture structural damage due to its high sensitivity.