Information Fusion and Finite Element Model Simulations for Bridge Condition Prognosis with Conflicting Data

Bridge inspection records are essential for engineers to diagnose structural conditions and make appropriate maintenance decisions. However, missing and conflicting data among multiple bridge inspection records documented at different times brings challenges to reliable condition assessments. Bridge engineers screen and validate the data from various inspection records and fuse validated data for condition assessments. Unfortunately, only experienced bridge engineers could achieve reliable information fusion. This paper presents an approach that automatically examines conflicting data from multiple inspection reports collected at different times. The proposed method uses finite element model simulations to identify how documented cracks could cause load redistributions across the structure and produce additional cracking areas. A "cracking location" voting strategy can help quantify the trustworthiness of the combinations of various crack locations for resolving data conflicts. The results indicate that the proposed method can fuse conflict bridge crack information from multiple inspection records into reliable bridge condition assessments.