Hybrid Time-Frequency Blind Source Separation towards Ambient System Identification of Structures
Abstract The last two decades have seen extraordinary growth in the development and practical implementation of structural control and monitoring technologies in many buildings and bridges world-wide. Central to the effective use of these technologies is our ability to identify the dynamic system properties under ambient loading conditions. In this presentation, the speaker will introduce a new class of system identification methods based on the principles of blind source separation (BSS). The focus of this presentation will be on specific tools of BSS that employ second-order statistics. Conventional BSS methods fail to perform adequately in noisy environments, with sparse measurements, and in the presence of low-energy and closely-spaced modes. In order to overcome these challenges, the speaker will present new methods that integrate time-frequency decomposition of signals with conventional second-order BSS methods. Results of an ongoing full-scale tower study in Canada will also be presented to illustrate the performance of the newly proposed methods. About the speaker Dr. Sriram Narasimhan currently serves as an Assistant Professor in the Civil and Environmental Engineering at the University of Waterloo in Canada. His main research areas are in structural control and identification of structural, mechanical and biological systems. Dr. Narasimhan received his PhD in 2005 from Rice University in Houston, Texas, USA. In addition to academia, Dr. Narasimhan has also spent over 5 years working in the consulting industry. He currently serves on the editorial boards of the ASCE Journal of Bridge Engineering and the Journal of Structural Control and Health Monitoring.