Statistical Time Series Methods forSHM

Abstract

Statistical time series methods for 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 includes fault detection, identification (localization), as well as magnitude estimation. The principles and operation of methods that utilize the time or frequency domains on a periodic inspection basis are explained, and the methods are classified into various categories under the general nonparametric and parametric classes. Representative methods from each category are outlined and their use is illustrated on a laboratory structure.

Statistical time series methods for damage diagnosis in a scale aircraft skeleton structure: loosened bolts damage scenarios

“…The method's characteristic quantity is the PSD function Q = S(ω) (ω designating frequency -table 3). Damage detection is based on confirmation of statistically significant deviations from the nominal healthy case at some frequency [1,2]. Damage identification may be achieved by performing hypothesis tests comparing the current PSD to those corresponding to different damage types and obtained in the baseline phase.…”

“…They offer a number of advantages, including no requirement for physics based or finite element models, no requirement for complete modal models, effective treatment of uncertainties, and statistical decision making with specified performance characteristics [1,2]. These methods form an important and useful category within the broader family of vibration based methods for SHM [3,4].…”

“…Damage diagnosis is then accomplished via statistical decision making consisting of comparing, in a statistical sense, the current characteristic quantity Q u with that of each potential state as determined in the baseline phase (inspection phase). For an overview of the basic principles and the main statistical time series methods for SHM the interested reader is referred to [1,2]. An experimental assessment of several methods is provided in [5].…”

“…An experimental assessment of several methods is provided in [5]. Non-parametric methods are those based on non-parametric time series representations, such as the spectral density function [1,2], and have received limited attention in the literature [5][6][7]. Parametric time series methods are those based on parametric time series representations, such as the AutoRegressive Moving Average (ARMA) models [1,2].…”

“…Non-parametric methods are those based on non-parametric time series representations, such as the spectral density function [1,2], and have received limited attention in the literature [5][6][7]. Parametric time series methods are those based on parametric time series representations, such as the AutoRegressive Moving Average (ARMA) models [1,2]. This latter category has attracted increased attention recently [5,[8][9][10][11], with approaches for accounting for varying environmental conditions being also considered [10,11].…”