Toogle Left

BATir banner

 

logo SMC logo SMC  DYNA logo SMC  DYNA  SHM SHM under changing environmental conditions
   

Contact

Arnaud Deraemaeker

   
Keywords structural health monitoring, environmental conditions, Mahalanobis squared distance, filtering
   

Collaborations

Prof K. Worden, University of Sheffield, UK

Prof J. Kullaa, Helsinki Metropolia University of Applied Sciences, Finland

   

Motivation

Vibration-based Structural Health Monitoring (SHM) techniques have been around for many years and are still today an active topic of research. Despite this fact, very few industrial applications exist. The main reason is the lack of robustness of these techniques with respect to changing environmental conditions.

Two major trends coexist in the field: model-based and data based techniques. Model based techniques are often sophisticated and require a high degree of engineering knowledge and more hardware and
software resources. They have however more potential to cover all levels of SHM, from damage detection to damage prognosis. On the other hand, data-based techniques are appealing because they are very simple and require less engineering knowledge as well as limited hardware and software. From that point of view, they are ideal candidates for industrial applications. These methods are however generally limited to the lowest levels of SHM: damage detection and in some cases, damage localisation. The three basic elements of data-based damage detection are (i) a permanent sensor network system, (ii) an automated procedure for real-time feature extraction, and (iii) a robust novelty detector.

It is well known in civil engineering that environmental conditions can cause changes in the measured features of the same order of magnitude or greater than the expected changes due to damage. This work is therefore focused on the development of a novelty detector robust against environmental conditions.

   

Changing environmental conditions

As an example, we consider the wooden bridge represented on Figure 1 (left). The eigenfrequencies and mode shapes have been measured continuously for a long period of time during which the changing temperature has been recorded. Figure 1(right) shows the evolution of  the sixth natural frequency of the bridge as a function of the temperature. For the last 200 samples, a structural modification is performed on the bridge through an added mass. One sees clearly that the effect of the structural change is much smaller than the effect of changing temperature on the eigenfrequency of the bridge

 

bridge frequency

Figure 1 : Effect of the temperature and the damage on the sixth eigenfrequency of a wooden bridge

   

Filtering environmental effects

The technique used in this study consists in filtering the effects of temperature without actually measuring it. For that, a training set is needed which covers the range of environmental changes the structure will undergo. The method consists in using the Mahalanobis squared distance with a covariance matrix [C] computed on this training set containing the data under changing environmental conditions. Figure 2 illustrates the evolution of the Mahalanobis squared distance as a function of the sample number and shows the efficiency of the filtering : the effect of structural change is clearly differenciated from the effect of environment so that damage detection under changing environmental conditions is now possible.

 

bridge mahalanobis

Figure 2 : Evolution of the Mahalanobis squared distance : the effect of a structural change is clearly differenciated from the effect of environment.

   

Selected publications

[1] A. Deraemaeker, E. Reynders, G. De Roeck, and J. Kullaa. Vibration-based structural health monitoring using output-only measurements under changing environment. Mechanical Systems and Signal Processing, 22:34–56, 2008.

[2] A. Deraemaeker and K. Worden. On the use of the mahalanobis squared-distance to filter out environmental effects in structural health monitoring. In Proc CSNDD 2014, Agadir, Morocco, May 2014.