Thermal Effect Identification and Bridge Damage Disclosure by using Blind Source Separation Method

Abstract

In this paper, blind source separation idea is employed to identify abnormal variations in time-history sensor measurements, which can be the indication of structural damage inside the bridge. Since the measurements are the mixed structural responses due to the complicated surroundings, environmental impact should not be ignored any more. The reason is environmental variations can not only change structural properties but also mask the response changes that indicate structural performance degradation. As it is difficult to procure the complex loading surroundings, the proposed method, thus, employs principal component analysis and independent component analysis method to uncover temperature variations and subsequently detect damage, without any prior information of the loading conditions and pre knowledge of structural physical models. The principal component analysis is employed as an assistance to reveal the number of latent loading conditions. Thereafter, the independent component analysis, one of the popular solution for blind separation stream, is applied on target sensor set to extract temperatureinduced variations and identify damage. The simulated damage scenarios are introduced into the aluminium bridge, which is a scaled model of a Japanese railway bridge. The results obtained from this case study illustrate that not only the temperature effects can be identified, but also anomalous variations can be uncovered along with their position.