Improved particle filter algorithm based on piezoelectric Lamb wave monitoring characteristics

Abstract

The prediction of fatigue crack growth by particle filter algorithm is to establish the state space-model of particle filter by using pairs formula combined with state noise model, to establish the observation space model of particle filter by using Lamb and its damage factor monitoring crack growth method combined with observation noise model, and to predict the crack growth by using the state-space model at the same time. In the process of monitoring fatigue crack growth by Lamb wave, the longer the crack is, the greater the uncertainty of the damage factor is. Based on the above characteristics, an observation space model with an observation noise model continuously modified with crack length is established. The hole edge crack propagation experiments are carried out to predict the crack propagation of particle filter algorithm and the improved particle filter algorithm. The experimental results show that the improved particle filter algorithm is not easy to diverge in the later stage of crack propagation, that is, after 25,000 cycles in this experiment. After 27650 cycles, the crack propagation prediction error is only 0.7mm.