Damage localization of long-span bridges based on stress influence lines and information fusion technique and its validation

To ensure bridge safety and functionality under in-service conditions, detecting local abnormalities of a long-span bridge at the early stage is always a desirable but challenging task. Compared with bridge global responses (such as displacement and acceleration), stress/strain can be more conveniently measured and is often more sensitive to local damages. Stress Influence Lines (SIL) or its derivatives are recognized as very promising indices for damage detection. This paper explores a novel damage localization approach by synthesizing SIL measurements from multiple locations, in which Dempster-Shafer data fusion technique is utilized. Compared with the measurement from a single sensor, more reliable damagerelated information with the improved sensitivity and capability in damage localization can be obtained by synthesizing the measured SILs from a number of sensors. The effectiveness of the proposed method is validated through a numerical case study of the Tsing Ma Suspension Bridge. Different hypothetical scenarios, including single-damage case, double-damage, and no-damage cases, are considered in the validation. After a comparison of damage detection results using information of single sensor, it is validated that the data fusion technique can effectively enhance the consistent information (e.g., damage-induced structural change) and minimize non-consistent information (such as “noise” effect) from multiple sensors installed around the damage. Then parameter analysis is conducted to examine the effect of sensor number. The result shows that the index of damage localization increases along with the number of sensors.