Novelty detection by nonlinear factor analysis for structural health monitoring

In vibration-based structural health monitoring damage in structure is tried to detect from damage-sensitive features. Because neither prior information nor data about expected damage are normally available, damage detection problem must be solved by using a novelty detection approach. Features, which are sensitive to damage, are often sensitive to environmental and operational variations. Therefore elimination of these variations is essential for reliable damage detection. At present many of the damage detection methods are linear, though it has been shown that many of the vibration changes in structures are bilinear or nonlinear. This paper proposes to use nonlinear factor analysis to detect damage via elimination of external effects from damage features. The effectiveness of the proposed method is demonstrated by analyzing the experimental Z24 Bridge data with a comparison to a linear method. It is shown that elimination of adverse effects and damage detection are feasible.

[1]  Hoon Sohn,et al.  Effects of environmental and operational variability on structural health monitoring , 2007, Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[2]  P. Guillaume,et al.  A robust singular value decomposition for damage detection under changing operating conditions and structural uncertainties , 2005 .

[3]  Hoon Sohn,et al.  VIBRATION-BASED DAMAGE DETECTION USING STATISTICAL PROCESS CONTROL , 2001 .

[4]  Charles R. Farrar,et al.  Novelty detection in a changing environment: Regression and interpolation approaches , 2002 .

[5]  Nikunj C. Oza,et al.  Online Ensemble Learning , 2000, AAAI/IAAI.

[6]  Bart Peeters,et al.  System identification and damage detection in civil engineering , 2000 .

[7]  Jaakko Hollmén,et al.  Feature Extraction and Selection from Vibration Measurements for Structural Health Monitoring , 2009, IDA.

[8]  Antti Honkela,et al.  Bayesian Non-Linear Independent Component Analysis by Multi-Layer Perceptrons , 2000 .

[9]  Cecilia Surace,et al.  Damage detection using Singular Value Decomposition , 1997 .

[10]  Antti Honkela,et al.  Unsupervised Variational Bayesian Learning of Nonlinear Models , 2004, NIPS.

[11]  Tapani Raiko,et al.  Tkk Reports in Information and Computer Science Practical Approaches to Principal Component Analysis in the Presence of Missing Values Tkk Reports in Information and Computer Science Practical Approaches to Principal Component Analysis in the Presence of Missing Values , 2022 .

[12]  Hoon Sohn,et al.  Statistical Damage Classification Under Changing Environmental and Operational Conditions , 2002 .

[13]  R. A. Leibler,et al.  On Information and Sufficiency , 1951 .

[14]  Gaëtan Kerschen,et al.  Structural damage diagnosis under varying environmental conditions—Part I: A linear analysis , 2005 .

[15]  Jyrki Kullaa,et al.  DAMAGE DETECTION OF THE Z24 BRIDGE USING CONTROL CHARTS , 2003 .

[16]  Neil D. Lawrence,et al.  Gaussian Process Latent Variable Models for Fault Detection , 2007, 2007 IEEE Symposium on Computational Intelligence and Data Mining.