On-line updating Gaussian mixture model for aircraft wing spar damage evaluation under time-varying boundary condition

Abstract

Structural health monitoring technology for aerospace structures has gradually turned from fundamental research to practical implementations. However, real aerospace structures work under time-varying conditions that introduce uncertainties to signal features that are extracted from sensor signals, giving rise to difficulty in reliably evaluating the damage. This paper proposes an online updating Gaussian Mixture Model (GMM)-based damage evaluation method to improve damage evaluation reliability under time-varying conditions. In this method, Lamb-wave-signal variation indexes and principle component analysis (PCA) are adopted to obtain the signal features. A baseline GMM is constructed on the signal features acquired under time-varying conditions when the structure is in a healthy state. By adopting the online updating mechanism based on a moving feature sample set and inner probability structural reconstruction, the probability structures of the GMM can be updated over time with new monitoring signal features to track the damage progress online continuously under time-varying conditions. This method can be implemented without any physical model of damage or structure. A real aircraft wing spar, which is an important load-bearing structure of an aircraft, is adopted to validate the proposed method. The validation results show that the method is effective for edge crack growth monitoring of the wing spar bolts holes under the time-varying changes in the tightness degree of the bolts.