Structure Damage Detection Based on Ensemble Learning

Structure damage detection is of great significance for structure maintenance, repair and rehabilitation soon after damage occurs. Data driven approaches to structure damage detection based on signal processing methods and Machine Learning algorithms have long been investigated. This paper proposed applications of Ensemble Learning to boost damage detection performance. Cross correlation function and wavelet packet decomposition are adopted to extract damage sensitive features from raw sensor data. Ensemble Learning algorithm such as Random Forest and XGBoost are used to train the damage pattern classifier. Results obtained in experiments validate Ensemble Learning algorithms' effectiveness in structure damage detection. This method can be used on practical structure damage detection problem in Aerospace Engineering.