Improving the performances of the autoregressive method in modal identification of output-only systems using Hilbert vibration decomposition method

Abstract

In this paper, an enhanced method for extraction of modal parameters (frequencies and damping ratios) of a structure from stationary or non-stationary measurement dynamic responses recorded by a sensor is presented. Surely, one of the simplest methods in area of ambient modal identification (operational modal analysis) is autoregressive method. Major problem of autoregressive method is that for identification of m modes of a structure, at least m sensors are needed. Besides, this method like other similar methods in this area such as frequency domain decomposition and stochastic subspace identification is appropriate for extraction of modal parameters from stationary measured dynamic structural responses. To address these issues, in this study, the Hilbert vibration decomposition method, which is a simple method for time-varying vibration decomposition based on the Hilbert transform, is adopted to improve the performance of the autoregressive method for extraction of frequencies and damping ratios of a structure from stationary or non-stationary responses recorded by a sensor. The efficiency and performance of the newly enhanced method are investigated through two numerical examples and a verification example. The first numerical example deals with a single-degree-of-freedom system subjected to a non-stationary force and the second one presents a two-degree-of-freedom structure excited by a stationary force. Finally, by using the proposed method, the frequencies and damping ratios of a support tower of the segmental bridge via an experimental test are obtained. The results indicated that the proposed method adequately estimated the frequencies and damping ratios of a structure from stationary and non-stationary responses recorded by only one sensor. Moreover, it is found that this method outperforms other relevant methods when dealing with non-stationary responses. Consequently, the enhanced method is strongly recommended for extraction of the frequencies and damping ratios of the structures from stationary or non-stationary responses, especially when the dynamic response of the structure is non-stationary and measured using only one sensor.

Keywords

• autoregressive method
• hilbert vibration decomposition
• modal identification
• non-stationary responses