Lamb Wave-based Monitoring of Fatigue Crack Propagation using Principal Component Regression

Abstract

Fatigue crack is an important factor affecting structural safety, and it is of great significance for accurate monitoring of fatigue crack propagation. This paper presents a Lamb Wave-based method for quantitative monitoring of fatigue crack propagation. In this method, various types of damage features are extracted in both time and frequency domains to comprehensively describe the Lamb wave changes. To address the problem of multicollinearity in damage features, principal component regression (PCR) is adopted to establish a quantitative model between damage features and crack size. The PCR model is established and validated by the experimental data of aluminum alloy plates. Experimental results reveal that the proposed PCR model is able to accurately monitor the fatigue crack propagation, and it performs far better than traditional multiple linear regression (MLR) model.