Data Fusion Technique for Bridge Safety Assessment

Abstract This article proposes a multilevel assessment system tailored to the health condition evaluation of prestressed continuous concrete bridges. The main structure of the assessment system is divided into a prior database and data processing unit. Some of the factors mainly influencing bridge safety, including carbonation, freeze-thaw cycles, steel corrosion, overloading, and temperature, are utilized to form various load cases as a basis for the prior database. To evaluate the bridge safety conditions comprehensively and scientifically, artificial intelligence methods and data fusion techniques based on information entropy, fuzzy analytical hierarchy, and the Dempster-Shafer theory are utilized to establish the data processing unit. An ambient excitation modal test was conducted to verify the numerical model. In situ mechanical responses from monitoring sensors and visual inspection results are fed back to the data processing unit as indicators of the safety grade of the bridge. The novel assessment system proposed here can feasibly and effectively provide support for bridge management and maintenance.