Abstract

Structural Health Monitoring (SHM) has drawn attention of many researchers recently. The reason is its huge effect on reduction in maintenance costs as well as increasing reliability of mechanical devices.
In this thesis the concept of SHM is explained and a damage detection methodology is proposed using Auto Regressive (AR) parameters for truss type structures.
The AR parameters of a healthy case are assumed to be the reference baseline data. A Damage Index is then defined to be the standard deviation of any other unknown signal from the baseline data. The proposed index provides an effective tool to detect the damage in the structure.
Sensor arrangement optimization has been performed as another part of this thesis which is a study on finding the optimum sensor arrangement to interrogate the most useful data given a limited number of sensors.
The localization process needs data classification techniques and has been conducted using Support Vector Machine (SVM) in this research for the first time. It is shown that SVM can successfully classify different signals that are extracted from a 3D sample truss structure. This accomplished through generating large sets of simulated data forwarded to SVM tool to construct a Meta model which further is used to predict the unknown signals and find the most correlated “known” category and reports its case label as the best match for the “unknown” signal.
At the end, an extensive sensitivity analysis has been performed to study the effect of parameter changes to the detection and localization processes.