Early Fatigue Crack Damage Identification by Multi-classification Support-Vector Machine Based on Lamb Wave and Temperature Compensation

Damage identification based on Lamb waves is one of the most effective technology for structural health monitoring and utilizes characteristic parameters to determine the degree of damage. However, the influence of temperature on these characteristic parameters is equivalent to the degree of damage, reducing the accuracy of damage identification. In this paper, a temperature compensation method for damage identification is proposed based on the extraction of Lamb wave characteristic parameters. Hilbert transform and Levenberg–Marquardt algorithm are employed to estimate the amplitude and phase parameters varying with temperature. The estimated parameters are applied to establish a temperature compensation model. The compensation model can reconstruct signals varying with temperature under different fatigue damage conditions and eliminate the influence of temperature on characteristic parameters. The response signals of random temperature under different damage conditions are selected for verification, and the reconstructed characteristic parameters are compared with the reference characteristic parameters. Besides, the SVM multi-classification algorithm is adopted to cross-validate the temperature compensation results. The results demonstrate that temperature compensation significantly improves classification accuracy.