Vida Ghasemi - Research Assistant, School of Civil Engineering, Iran University of Science and Technology, Tehran, Iran
Fereidoun Amini - Professor, School of Civil Engineering, Iran University of Science and Technology, Tehran, Iran
Elham Rajabi - Postdoctoral Fellow, Center of Excellence for Fundamental Studies in Structural Engineering, School of Civil Engineering, Iran University of Science and Technology, Iran

This paper presents an adaptive blind source separation technique, natural gradient algorithm (NGA) to identify modal characteristics of structures in real-time. To do this, at first, the NGA algorithm is used to estimate modal matrix and modal coordinates online, then modal parameters extracted from modal coordinates. The NGA updates the un-mixing matrix for each step. With characteristics such as the need for little data, lower computational cost, small storage capacity and lower analysis time the NGA algorithm is more suitable in civil engineering applications, especially in online structural identification. The efficiency of the proposed method is investigated with synthetic example

Adaptive blind sources separation, Modal identification, Natural gradient algorithm, On-line structural identification, Output-only system identification