On Selection of Data Fusion Schemes for Structural Damage Evaluation

Data fusion plays a pivotal role to achieve reasonable accuracy and precision in identifying structural damage. An appropriate fusion process can reduce imprecision, uncertainties and incompleteness, therefore increasing the robustness and reliability of identification. The present work compared three major fusion schemes, i.e., disjunctive, conjunctive, and compromise fusion, in terms of their effectiveness to estimate mono- and multi-delamination in carbon fiber-epoxy composite structures. (1) Time-of-flight was extracted from Lamb wave signals rendered by an active sensor network, to attain the loci of locations of all possible damage instance(s) in the structure under inspection, which served as the prior perceptions of sensors as to the areas with possibility of damage occurrence; and (2) the entire structure was virtually meshed and the prior perceptions of individual sensors were further quantified at each spatial mesh node using the distance between nodes and all loci established from (1), to form prior probabilities of damage occurrence at nodes. Then, three fusion schemes were employed to fuse the prior probabilities at all spatial nodes to shape a posterior consensus concerning the overall health status of the structure. Hybrid fusion by combining three basic schemes was also explored. Conclusions drawn from this study have given an indication on how to select a better, if not the most optimal, data fusion scheme for structural damage evaluation.