Interpreting structural information embedded in large-scale monitoring data is the main task for structural health monitoring studies. Most existing research investigates different types of structural responses separately and evaluates their individual effects on structural condition without integrated considerations. This single-class approach may overlook certain degradation patterns. Several studies that employ fused monitoring data for structural condition assessment generally lack consideration of the influence of environmental factors. To address this limitation, this study proposes a temperature-informed monitoring data fusion method and the corresponding structural condition assessment framework. The main novelty of this study lies in redefining the settings of the monitoring data fusion process. By embedding temperature information into the fusion process as the basis for probabilistic evidence construction, the proposed framework improves the robustness of the fused condition assessment. First, the sensitivity of structural thermal effects to damage was theoretically analyzed. Gaussian process (GP) models were then employed to describe the nonlinear relationships between the temperature and structural responses. Structural changes were quantified by the deviations between the GP models fitted to data from different periods. Subsequently, using evidence reasoning theory, these deviations from multiple sensor channels were treated as pieces of evidence reflecting structural status. By accounting for the correlations among evidence sources and inherent uncertainties in the reasoning process, a probabilistic fusion of multiple types of data was achieved. The proposed method was validated using monitoring data from a strengthened continuous viaduct bridge. Results indicate that the framework effectively captures the bridge’s long-term condition and a comparative analysis clearly demonstrates its superiority over assessments based on a single-class of structural responses.
Gong et al. (Sat,) studied this question.