Lead University: Lehigh University
PI: Shamim Pakzad, Department of Civil and Environmental Engineering
PA Industry: Specialty Engineering, Inc.
Applications of dense sensor networks in Structural Health Monitoring (SHM) projects inevitably impose high computational expense to the SHM algorithms in processing such large monitoring data sets. These necessitate the design of SHM frameworks that are scalable to the size of the network in terms of their computational complexity. The objective of this proposal is to develop and evaluate damage detection algorithms enable to infer damage correctly by analyzing a subset sample of the collected data.
In the proposed compressive damage localization model, first a set of sensors from the network are randomly sampled. Measurement from these sampled sensors are processed to extract damage sensitive features. These features undergo statistical change detection tests and classify the network into “damaged” or ”not damaged” regions. This classification directs the new sampling boundary and is sequentially updated as new sensors are added to processing subset and more information about location of damage is provided. This research also aims to compare suitability of several damage sensitive features, change point analysis, and data compression methods for use in the context of compact damage detection and find those capable of lowering the computational expense of search for location of damage and improve the scalability of the damage localization algorithms.
Performance of the proposed algorithm is evaluated on gusset plate connections. Pre- and post-damage strain distributions in the gusset plate are used for damage diagnosis. These evaluation studies are implemented on damage scenarios in finite element simulations as well as experimental set up where Digital Image Correlation is used to collect a highly dense grid of the strain field.