Time series-based SHM using PCA with application to ASCE benchmark structure

• Published in: Journal of civil structural health monitoring Vol. 10; no. 5; pp. 899 - 911
• Main Authors: Kumar, Kundan, Biswas, Prabir Kumar, Dhang, Nirjhar
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2020; Springer Nature B.V
• Subjects: Algorithms; Autoregressive models; Benchmarks; Catastrophic failure analysis; Civil Engineering; Computer networks; Control; Corrosion fatigue; Damage detection; Data transmission; Dynamical Systems; Engineering; Measurement Science and Instrumentation; Natural disasters; Noise measurement; Original Paper; Principal components analysis; Remote monitoring; Structural damage; Structural health monitoring; Time series; Transmission rate (communications); Vibration; Wireless sensor networks; Mahalanobis squared distance; Outlier detection; Eigenspace; Structural health monitoring; Damage detection; Principal component analysis
• ISSN: 2190-5452; 2190-5479
• DOI: 10.1007/s13349-020-00423-2

Detecting damage at an early stage can avoid a serious catastrophic failure of structures due to inevitable cause, such as fatigue, environmental corrosion, and natural disasters. Various damage detection algorithms have been proposed based on autoregressive model using time series data, which are computationally expensive, and the selection of an optimal order of the model requires extra expertise. In this paper, computationally efficient algorithm is proposed to process the time series data using principal component analysis (PCA) in an effective way. PCA is utilized to model a feature space to compute damage sensitive features insensitive to environmental variations and measurement noise. The modeled feature space preserves the damage information along with eliminates the consequences of environmental variations and measurement noise. Furthermore, Mahalanobis squared distance is adopted to compute damage index as the severity of the damage. The proposed method is validated on analytical models of IASC–ASCE benchmark structure. The test results show that the proposed damage diagnosis method can be useful for wireless sensor network-based structural health monitoring with less computation and low data transmission rate.