Dual crack growth prognosis by using a mixture proposal particle filter and on-line crack monitoring

On-line prognosis of fatigue cracks in the structure is challenging due to various uncertainties affecting fatigue crack initiation and growth. This paper proposes an on-line prognosis strategy for fatigue cracks by incorporating the mixture proposal particle filter (MPPF) and structural health monitoring (SHM) results. In this method, a dynamic crack evolution model is proposed to deal with the situation that more than one crack occurs and grows in the structure. Meanwhile, crack sizes monitored by the SHM technique are incorporated to construct an effective mixture proposal of the importance probability density, which is the key for sampling new particles. Further, posterior estimations of the fatigue crack sizes and the crack evolution model parameters are evaluated with these particles, based on which the prognosis of fatigue crack growth is carried out. A leave-one-out validation is performed on the dual crack growth problem of the hole-edge-cracked structure, demonstrating the effectiveness of the proposed method.