Published: 30 September 2012

Experimental study for structural damage identification with incomplete measurements

Rui Li1
Li Zhou2
Jann N. Yang3
Views 1
Reads 0
Downloads 1073

Abstract

Finite element formulation for the equation of motion should be used for identification of local damage of a structure at the element level. However, the finite element model of a structure involves a large number of degrees of freedom and requires a large number of sensor measurements. To avoid measuring vibration responses, which are difficult to obtain (e.g. rotational accelerations), and to reduce the required number of sensors, a reduced-order finite element formulation along with the adaptive sequential nonlinear least square estimation technique is proposed in this paper to identify local damages of structures. To verify the applicability and effectiveness of the proposed approach, two series of damage detection experiments were conducted using scaled cantilever beams. One series of experimental tests were conducted for the detection of constant damages. In this test series, different damage severities were simulated by drilling different number of circular holes with different sizes in a particular element of a cantilever beam. Another series of experimental tests were conducted to verify the online damage tracking capability of the proposed approach. In this test series, a stiffness element device was installed in a particular element of another cantilever beam to simulate the abrupt stiffness reduction of that element during the test. Experimental results demonstrate that the proposed reduced-order finite element model along with the adaptive sequential nonlinear least square estimation technique is effective and accurate in detection of structural damages, including the damage location and severity using only a limited number of sensors.

About this article

Received
30 May 2012
Accepted
04 September 2012
Published
30 September 2012
Keywords
experimental study
damage detection
structural health monitoring
system identification
adaptive sequential nonlinear least square estimation