Experimental incremental system identification method using separate time windows on basis of ambient signals

Abstract

Experimental system identification methods, such as Frequency Domain Decomposition (FDD), Autoregressive Moving Average Model (ARMA), Autoregressive Models with eXogenous input (ARX), Kalman Filter and Stochastic Subspace Identification (SSI), are commonly used in civil engineering to determine dynamic parameters of existing structures. Basis for these methods are in-situ measurements, which can be very time-consuming and cost-intensive depending on the complexity of the structure. This paper investigates the possibility to reduce the in-situ measurement effort by introducing a new method, which bases on incremental measurements by using only a single sensor in separate time windows. The proposed incremental identification method (IIM) requires stationary ergodic response signals of the structure induced by ambient vibrations with white noise density. Therefore, after each incremental measurement a quality-check of the response signal should be conducted to verify the applicability of the theory. This approach ensures the comparability of the input signals with each other and thus the reproducibility of the identified dynamic behavior. For this purpose, a signal evaluation criterion is defined. For low-quality data, which cannot satisfy this criterion, special signal processing methods have to be applied. With the signals, which already accomplish the evaluation criterion, the identification of the system parameter can then be carried out by using one of the above mentioned system identification methods, such as FDD. The IIM is applied so far both on numerical and experimental examples. In this paper the validation of the IIM is reached by identifying the parameters of the IASC-ASCE Structural Health Monitoring Benchmark Problem for different ambient simulated input signals.