Home>Research>Structural Health Monitoring

Damage Identification Methods For Wireless Sensors

Background on Wireless Sensing Devices

Leverage on Wireless Sensing Devices

With recent developments in wireless sensing devices, it will soon be possible to install numerous low-cost sensors on a single structure. This would enable better local damage identification.

Their advantages:

  • potentially low-cost
  • no need to put wires (wires have a high installation and maintenance cost)
  • can perform real-time computation

Methods that efficiently use wireless sensing devices for damage detection don't yet exist or are under development. We therefore focus our research on leveraging on wireless sensor platforms for damage identification.

Their Requirements

The main limitation of wireless sensing devices is energy. For autonomy, they usually operate on batteries or use energy harvesting devices that can only provide a limited amount of energy. Therefore, energy should be used with care. The two main energy sinks are computations and radio communications.


It consumes a lot of energy to use the radio of wireless sensing devices to communicate data over the network. Therefore, to keep the network scalable, it is needed to perform some sort of data compression or analysis before sending. In this way, communications are kept at a minimum.


Finally, wireless sensing devices have an on-board CPU, and can execute damage identification algorithms. Because this computational power is limited, special care should be taken to design of those algorithms.

Autoregressive Models

We are working on decentralized algorithms to detect local damage in civil structures. Sensors (accelerometers) are distributed over the structure, and are grouped in clusters, like in Figure 1. For each cluster, local models are built using data from the healthy structure. New measurements are then checked against those healthy models.

Multivariate Autoregressive Models for Local Damage Detection Using Small Clusters of Wireless Sensors (PDF 226KB, EWSHM, July 2006)

Wavelet Analysis

(in construction)