Abstract: Damage detection of the aerospace structures is a critical issue, during which the data from multiple types of sensors are redundant and inconsistency to each other. Rough set reduction is an effective tool for data mining and classification, which can be used for the above problems. Firstly, the damage features arc extracted from the multi-sensor information(e. g. displacement, acceleration or strain sensor) and modal parameters. Then, the rough set reduction techniquc is employed to obtain the core set of all features. Finally, the structural damage conditions arc identified through probabilistic neural network. hhe numerical simulation of a helicopter demonstrates that the rough set reduction technictue can not only decrease the spatial dimension of feature attributions,but also improve the identification accuracy.

Key words: Rough set reduction,Spacecraft,Fault diagnosis