Prognostic Algorithms for Flaw Growth Prediction in an Aircraft Wing

Hybrid resampling scheme for particle filter‐based inversion

“…, where f n is the state transition process, h n o is the measurement process, v n is the process noise, and μ n o is the measurement noise [27]. We considered a polynomial measurement model as in [15] to relate x n with z n measurement:…”

“…, where ′c′ is the polynomial coefficient which is obtained from the training database. The computational complexity of this measurement model is the Bth order, which is low [15]. These state transition and measurement models are the representation of a tracking problem [40,41].…”

“…Crack/flaw growth prediction can be made through prognostic algorithms [1,15]. The flaw growth in microstructures is typically a stochastic process [15].…”

“…Crack/flaw growth prediction can be made through prognostic algorithms [1,15]. The flaw growth in microstructures is typically a stochastic process [15]. The most common Bayesian technique to deal with stochastic data more efficaciously is a Kalman filter [15].…”

“…Alternatively, physical model-based iterative methods were also proposed that minimises the difference between actual and estimated measurements [7][8][9]. Specifically, conventional numerical model-based electromagnetic NDE signal inversion schemes are in practice [3,6,[9][10][11][12][13][14], but such techniques are computationally expensive [15]. Dogandzic reports Bayesian techniques for the detection and the estimation of defects from NDE signals [16,17].…”