Statistical Time Series Methods for Vibration Based Structural Health Monitoring
Statistical time series methods for vibration based structural health monitoring utilize random excitation and/or vibration response signals, statistical model building, and statistical decision making for inferring the health state of a structure. This i
- PDF / 1,739,561 Bytes
- 56 Pages / 439.37 x 666.142 pts Page_size
- 85 Downloads / 271 Views
1
Introduction
Structural Health Monitoring (SHM) involves the continual or continuous over time monitoring of a structure based on proper sensors which provide dynamic structural responses and other related data (such as environmental conditions), signal/data processing and analysis, as well as proper decision making for inferring the current health state of the structure. Once set up, an SHM procedure is ideally intended to be global (in the sense of covering the whole structure or a large part of it), automated, without necessitating human interaction, cost-effective, and capable of effectively treating the level I, II and III subproblems (Rytter, 1993), that is damage detection (simply detecting damage presence), damage identification (identifying the damage type/nature and location) and damage quantification (estimating the damage extent) – see section 2.2 for details. Historically, SHM may be thought of as an evolution of classical NonDestructive Testing (NDT) procedures (commonly based on ultrasound, W. Ostachowicz, J. Güemes (Eds.), New Trends in Structural Health Monitoring, CISM International Centre for Mechanical Sciences, DOI 10.1007/978-3-7091-1390-5_4, © CISM, Udine 2013
210
S. D. Fassois and F. P. Kopsaftopoulos
acoustic, radiography, eddy current, and thermal field principles – Doherty 1987; Doebling et al. 1996, 1998; Farrar et al. 2001; Balageas et al. 2006). NDT is however different, in that it is exercised on demand – usually on a periodic basis – without permanent sensors mounted on the structure and not necessarily in an automated fashion. NDT typically works locally, requiring access to the vicinity of the suspected damage location, while the procedure is often time consuming and costly. On the other hand, the SHM philosophy and principles are much closer to the general theory of fault diagnosis (see Basseville and Nikiforov 1993; Rytter 1993; Doebling et al. 1996, 1998; Natke and Cempel 1997; Salawu 1997; Farrar et al. 2001). Vibration based SHM is quite popular, as vibration is naturally available for many structures (aircraft, railway vehicles, bridges, and so on), while the technology for the precise measurement and processing of vibration signals has been available for a long time. For overviews of general vibration based methods see Doebling et al. (1996, 1998); Salawu (1997); Zou et al. (2000); Farrar et al. (2001); Sohn et al. (2003b); DeRoeck (2003); Carden and Fanning (2004); Montalv˜ ao et al. (2006). Also see Staszewski et al. (2004); Inman et al. (2005); Balageas et al. (2006); Fritzen (2006); Adams (2007). Statistical time series SHM methods form an important and rapidly evolving class within the broader context of vibration based SHM. Their three main elements are: (i) random excitation and/or vibration response signals (referred to as time series), (ii) statistical model building, and (iii) statistical decision making for inferring the health state of a structure. As with all vibration based methods, the fundamental principle upon which they are founded is that small ch
Data Loading...
