V. H. Vu

Short-Time Autoregressive (STAR) Modeling for Operational Modal Analysis of Non-stationary Vibration

In this chapter, a method based on an autoregressive model in a short-time scheme is developed for the modal analysis of vibrating structures whose properties may vary with time and is called Short-Time AutoRegressive (STAR) method. This new method allows for the successful modeling and identification of an output-only modal analysis. The originality of the proposed method lies in its specific handling of non-stationary vibrations, which allows the tracking of modal parameter changes in time. This chapter presents an update of the model with respect to model order and a noise-to-signal based criterion for the selection of the minimum model order. A length equal to four times the period of the lowest natural frequency has been numerically found to be efficient for the data block size and may be recommended for experimental applications. To validate the method, a system with three degrees of freedom is first simulated under a random excitation, and both stationary and non-stationary vibrations are considered. The method is finally applied on the real multichannel data measured on an experimental steel plate emerging from water, and is compared to the conventional Short-Time Fourier Transform (STFT) method. It is shown that the proposed method outperforms in terms of frequency identification, whatever the non-stationary behaviour (either slow or abrupt change) due to the added mass effect of the fluid.

Extraction of modal parameters for identification of time-varying systems using data-driven stochastic subspace identification

This paper presents a method for the extraction of modal parameters for identification of time-varying systems using Data-Driven Stochastic Subspace Identification (SSI-DATA). In practical applications of SSI-DATA, both the modal parameters and computational ones are mixed together in the identified results. In order to differentiate the structural ones from computational ones, a new method based on the eigen-decomposition of the state matrix constructed in SSI-DATA is proposed. The efficiency of the proposed method is demonstrated through numerical simulation of a lumped-mass system and experimental test of a moving robot for extracting excited natural frequencies of the system.

Modal analysis of a light-weight robot with a rotating tool installed at the end effector

This paper investigates vibration of a moving flexible robot through modal analysis and by constructing vibration spectra of operational signals. A vector autoregressive model combined with a sliding window technique is used for signal processing in order to take into account system nonstationarity. Modal decomposition is conducted on the state matrix constructed from the appropriate vector autoregressive model parameters. A complete modal decomposition and spectrum construction algorithm able of highlighting the structural modes and harmonic excitations is presented. Through accurate identification from the vector autoregressive model, the method presented is able to discriminate, display and monitor the harmonics and structural modes during the processes investigated. This method is validated first by numerical simulation and then experimentally with a flexible robot performing three processes: moving a manipulator through the workspace, steady rotation of a grinder on the end effector and moving the manipulator combined with rotating the grinder. It is found on the operating robot that participation of the first structural mode is negligible when rotating the grinder but must be taken into account when moving the manipulator. The analysis presented and results obtained provide a sound basis for further investigation of vibroimpact behaviour in a robotic grinding process.