Auto-Regressive Integrated Moving-Average Machine Learning for Damage Identification of Steel Frames

Auto-regressive (AR) time series (TS) models are useful for structural damage detection in vibration-based health monitoring (SHM). However, certain limitations, e.g., non-stationarity and subjective feature selection, have reduced its wide-spread use. With increasing trends machine learning (ML) technologies, automated recognition is becoming popular attracting many researchers. In this paper, we combined TS modeling ML classification to automatically extract features overcome the limitation of non-stationarity. We propose a two-stage framework, namely auto-regressive integrated moving-average (ARIMA-ML) with modules pre-processing, model parameter determination, extraction, classification. Based on shaking table tests space steel frame, floor acceleration data were collected labeled according experimental observations records. Subsequently, designed three tasks for: (1) global detection, (2) local (3) pattern recognition. The results from these indicated robustness accuracy proposed framework where 97%, 98%, 80% average segment achieved, respectively. confusion matrix showed unbiased performance even under an imbalanced-class distribution. summary, presented study revealed high potential ARIMA-ML SHM.