Section: New Results

Damage diagnosis

Damage detection by perturbation analysis and additive change detection theory

Participants : Michael Doehler, Laurent Mevel, Qinghua Zhang.

The monitoring of mechanical systems aims at detecting damages at an early stage, in general by using output-only vibration measurements under ambient excitation. In this paper, a method is proposed for the detection and isolation of small changes in the physical parameters of a linear mechanical system. Based on a recent work where the multiplicative change detection problem is transformed to an additive one by means of perturbation analysis, changes in the eigenvalues and eigenvectors of the mechanical system are considered in the first step. In a second step, these changes are related to physical parameters of the mechanical system. Finally, another transformation further simplifies the detection and isolation problem into the framework of a linear regression subject to additive white Gaussian noises, leading to a numerically efficient solution of the considered problems. A numerical example of a simulated mechanical structure is reported for damage detection and localization [31].

Damage localization using the statistical subspace damage localization method

Participants : Michael Doehler, Laurent Mevel, Saeid Allahdadian.

This work is happening during a thesis in collaboration with C. Ventura at UBC, Vancouver.

In this paper the statistical subspace damage localization (SSDL) method is employed in localizing the damage in a real structure, namely the Yellow frame. The SSDL method is developed for real testing conditions and tested in two damage configurations. It was demonstrated that the SSDL method can localize the damage robustly in the Yellow frame for simple and multiple distinct damage scenarios using the analytical modal parameters. The method is described and its effectiveness is demonstrated [24].

Stochastic Subspace-Based Damage Detection with Uncertainty in the Reference Null Space

Participants : Michael Doehler, Laurent Mevel, Eva Viefhues.

This work is happening during a thesis in collaboration with F. Hille at BAM, Berlin.

This paper deals with uncertainty considerations in damage diagnosis using the stochastic subspace-based damage detection technique. With this method, a model is estimated from data in a (healthy) reference state and confronted to measurement data from the possibly damaged state in a hypothesis test. Previously, only the uncertainty related to the measurement data was considered in this test, whereas the uncertainty in the estimation of the reference model has not been considered. We derive a new test framework, which takes into account both the uncertainties in the estimation of the reference model as well as the uncertainties related to the measurement data. Perturbation theory is applied to obtain the relevant covariances. In a numerical study the effect of the new computation is shown, when the reference model is estimated with different accuracies, and the performance of the hypothesis tests is evaluated for small damages. Using the derived covariance scheme increases the probability of detection when the reference model estimate is subject to high uncertainty, leading to a more reliable test [41].

Statistical damage localization with stochastic load vectors

Participants : Md Delwar Hossain Bhuyan, Michael Doehler, Laurent Mevel, Guillaume Gautier.

This work is in collaboration with F. Schoefs and Y. Lecieux, GEM, Nantes.

The Stochastic Dynamic Damage Locating Vector (SDDLV) method is a damage localization method based on both a Finite Element (FE) model of the structure and modal parameters estimated from measurements in the damage and reference states of the system. A vector is obtained in the null space of the changes in the transfer matrix from both states and then applied as a load vector to the model. The damage location is related to this stress where it is close to zero. An important theoretical limitation was that the number of modes used in the computation could not be higher than the number of sensors located on the structure. In this paper, the SDDLV method has been extended with a joint statistical approach for multiple mode sets, overcoming this restriction on the number of modes. Another problem is that the performance of the method can change considerably depending of the Laplace variable where the transfer function is evaluated. Particular attention is given to this choice and how to optimize it. The new approach is validated in numerical simulations and on experimental data. From these results, it can be seen that the success rate of finding the correct damage localization is increased when using multiple mode sets instead of a single mode set [15], [52], [27].

Transfer matrices-based statistical damage localization and quantification

Participants : Md Delwar Hossain Bhuyan, Michael Doehler, Laurent Mevel, Guillaume Gautier.

This work is in collaboration with GEM, Nantes and C. Ventura at UBC, Nantes.

Vibration measurements and a finite element model are used to locate loss of stiffness in a steel frame structure at the University of British Columbia. The Stochastic Dynamic Damage Locating Vector (SDDLV) is compared to a sensitivity based approach developed by the authors. Both approaches have in common to be built on the estimated transfer matrix difference between reference and damaged states. Both methods are tested for localization and quantification on a structure at University of British Columbia [26], [28].

Statistical damage localization based on Mahalanobis distance

Participant : Michael Doehler.

This work is in collaboration with Aalborg University, Structural Vibration Solutions and Universal Foundation in Denmark during the thesis of S. Gres (Aalborg University).

In this paper, a new Mahalanobis distance-based damage detection method is studied and compared to the wellknown subspace-based damage detection algorithm. Methods are implemented using control charts to enhance the resolution of the damage detection. The damage indicators are evaluated based on the ambient vibration signals from numerical simulations on a novel offshore support structure and experimental example of a full scale bridge. The results reveal that the performance of the two damage detection methods is similar, hereby implying merit of the new Mahalanobis distance-based approach, as it is less computationaly complex [32].

On the value of Information for SHM

Participant : Michael Doehler.

This work is issued from the COST Action TU1402.

The concept of value of information (VoI) enables quantification of the benefits provided by structural health monitoring (SHM) systems in principle. Its implementation is challenging, as it requires an explicit modelling of the structural system's life cycle, in particular of the decisions that are taken based on the SHM information. In this paper, we approach the VoI analysis through an influence diagram (ID), which supports the modelling process. We provide a simple example for illustration and discuss challenges associated with real-life implementation [39].

Structural system reliability and damage detection information

Participant : Michael Doehler.

This work is in collaboration with S. Thöns (DTU) during the thesis of L. Long (BAM).

This paper addresses the quantification of the value of damage detection system and algorithm information on the basis of Value of Information (VoI) analysis to enhance the benefit of damage detection information by providing the basis for its optimization before it is performed and implemented. The approach of the quantification the value of damage detection information builds upon the Bayesian decision theory facilitating the utilization of damage detection performance models, which describe the information and its precision on structural system level, facilitating actions to ensure the structural integrity and facilitating to describe the structural system performance and its functionality throughout the service life. The structural system performance is described with its functionality, its deterioration and its behavior under extreme loading. The structural system reliability given the damage detection information is determined utilizing Bayesian updating. The damage detection performance is described with the probability of indication for different component and system damage states taking into account type 1 and type 2 errors. The value of damage detection information is then calculated as the difference between the expected benefits and risks utilizing the damage detection information or not. With an application example of the developed approach based on a deteriorating Pratt truss system, the value of damage detection information is determined, demonstrating the potential of risk reduction and expected cost reduction [36].

Estimation of a cable resistance profile with readaptation of mismatched measurement instrument

Participants : Nassif Berrabah, Qinghua Zhang.

As the cumulative length of electric cables in modern systems is growing and as these systems age, it becomes of crucial importance to develop efficient tools to monitor the condition of wired connections. Therein, in contrast to hard faults (open or short circuits), the diagnosis of soft-faults requires a particular effort. Indeed, these faults are more difficult to detect, yet they are sometimes early warning signs of more important failures. In a previous paper, we proposed a method to compute the resistance profile of a cable from reflectometry measurements made at both ends of the cable. It enables detection, localization and estimation of dissipative soft-faults. In this reported work, we address the problem of impedance mismatch between the measurement instrument and the cable, based on a pre-processing of the measured data before running the estimation computations. It aims at reducing the impedance mismatch between instrumentation and the cable under test without physical intervention on the test fixtures. In addition, a measurement procedure has been developed in order to get the two-ends reflectometry measurements without actually connecting both ends of the cable under test to a single instrument [25].