Weighted adaptive Kalman filtering-based diverse information fusion for hole edge crack monitoring

Abstract

The information fusion of different sensors is an effective means of improving the damage quantification of aircraft structures operating in a complex environment. A weighted adaptive Kalman filtering-based information fusion method (WAKFIFM) is proposed to achieve the feature-level fusion of diverse signals of the hybrid piezoelectric-fiber optic sensor network. The damage identification results obtained from guided wave signals and strain signals are described by the quantification curves of damage index (DI) and crack length. A new fusion curve is obtained by fitting the above two quantification curves using the quadratic polynomial method. The mean square error and the root mean square error of the fusion curve and the actual curve are used as the error and the covariance of the proposed WAKFIFM, respectively. The weighted value in WAKFIFM is the reciprocal of the square of the actual measured maximum DI. The hole edge crack propagation experiments are conducted on both an aluminum plate and a carbon fiber reinforced polymer plate in a load-temperature changing environment.A new damage location method integrated with DI information is developed to obtain a more accurate result. The feature-level fusion results show that the proposed WAKFIFM can improve the accuracy of damage quantification, compared against that obtained from the single guided wave signals or fiber optic sensor signals.