Smartphone Application for Structural Health Monitoring of Bridges · Figueiredo E, Moldovan I, Alves P et al.

5. Conclusions

Smartphone-based applications for SHM still face technical challenges because of operator-induced uncertainties during data acquisition and feature extraction, a lack of control over the embedded sensing technology, and relatively low sensitivity and resolution of their built-in sensors. However, the end-user technology provided by smartphones has the potential to transform the way we perform bridge inspections, especially after catastrophic events such as earthquakes and tornados.

As a customized SHM process for damage detection, App4SHM does not intend to replace the visual inspections, rather it intends to provide more accurate assessment during bridge inspections and, at most, to reduce their frequency.

App4SHM was tested on data sets from a laboratory steel beam and two twin real-world concrete bridges. The former was designed to understand its sensing and damage detection capabilities in a more controlled environment, and the latter for damage detection in real world scenarios.

From the laboratory experiments, a high correlation between time series and, particularly, power spectral densities from different sensing systems (smartphone and two conventional accelerometers) was found. Comparing the average frequencies obtained resorting to the smartphone with those determined using the conventional sensing system, a maximum difference of 0.54% was obtained. The lab beam was also used to assess the sensitivity of App4SHM to damage, simulated through the addition of three different weights at two locations on the structure. It was concluded that damage becomes detectable when at least one of the monitored frequencies differs in more than two standard variations from its mean value under undamaged conditions.

The analyses conducted on two twin concrete bridges show that in terms of feature extraction, App4SHM outputs natural frequencies that are consistent with those obtained using conventional accelerometers (differences < 2.42%), even using short time-series durations, and within the variability observed with the traditional sensing system. Moreover, App4SHM correctly identifies the existence of outliers when observations from one of the bridges are fed into machine learning algorithms trained with data from the other (structurally similar) bridge.

The results from the two case studies show that App4SHM is reliable for both vibration measurement and damage detection in the context of SHM. However, as is the case with all individual components of the SHM, App4SHM is not meant to be used unaccompanied, but integrated into a Bridge Management System, along with complementary sensors and NDE.

The app and the sensing technology of smartphones still have some limitations. However, the laboratory and field tests have shown certain applications where we can obtain a reasonable trade-off between the potential lack of sensitivity and the advantages provided by the ease of installation and data acquisition as well as wireless transmission.