Research On Lamb Wave Modeling With Spectral Element Method And Multi-site Damage Quantitative Monitoring Algorithm

Lamb wave and piezoelectricity sensor based damage diagnosis method is sensitive to smalldamage and can realize regional monitoring that the sensor is not required to be placed where damageoccurs. Thus it has been the research hot spot of Structural Health Monitoring (SHM). Aiming ataeronautical structures, the thesis focuses on dynamical simulation of stress wave and quantitativemulti-damage monitoring for large-scale composite structures. The main creative jobs are as follows:(1) A plain strain numerical calculation model for the group velocity research of Lamb wave isproposed. The conclusion is verified that SEM has the advantage of less group velocity error andconsumed time over conventional FEM in the condition of same node density. The influence of groupvelocity accuracy using SEM by the shape function polynomial order and node density is researched.At last the benchmark how to choose related parameters to obtain a certain accuracy of group velocityis given.(2) A SEM dynamical simulation model of3D complex plane structures is proposed withABAQUS plane meshing and Serendipity2nd order element meshing optimization. By setting nodenumber of the element in3directions, the calculation accuracy of the model is controlled. Experimentresults on3different aluminum plates are compared with simulation results, which validates thecorrectness of the3D SEM dynamical simulation model.(3) A multi-damage quantitative dual merging monitoring algorithm is proposed. The algorithmmerges the features of both the damage index method and imaging method. The former is highlyefficient and the latter is capable of extracting damage feature comprehensively. Thus the mergingmonitoring algorithm quickly realizes multi-damage identification, imaging, localization and severitylevel estimation for large-scale composite. A complex carbon fiber composite wing box of aircraft isused for validation of the proposed algorithm. The experiment result shows that it correctly identifiesmultiple damages and the localization errors are all below25mm when the set group velocitydeviates at a range of10%from the calculated average group velocity. The estimated damage sizelevels are matched with the actual damage sizes with probabilities over90%.