Abstract Wind turbine blade integrity is critical to ensuring operational efficiency and extending equipment lifespan. However, existing monitoring methods struggle with the nonlinear and non-stationary nature of acoustic emission (AE) signals, as well as the presence of significant environmental noise. This study introduces an intelligent damage monitoring approach, Dual Phase Space-Mapped Symplectic Geometric Mode Decomposition (DS-SGMD), to enhance AE signal analysis in complex conditions. The proposed method improves phase space reconstruction by adaptively optimizing time delay parameters and selecting practical components based on geometric features of phase space trajectories. A pencil lead break test was used to establish a quantitative model linking AE signal amplitude and propagation distance, identifying a maximum adequate sensor spacing of 1400 mm. Simulation results demonstrate that DS-SGMD achieves a correlation coefficient of 0.8981 with reference damage signals and improves the signal-to-noise ratio by 6.26 dB. On a scaled wind turbine blade platform, DS-SGMD-based deep learning models achieved recognition accuracies of 98.8%, 98.1%, and 100% for intact, cracked, and damaged states, respectively, and improved delamination detection from 84.9% to 100%. Five-fold cross-validation yielded a mean accuracy of 98.6% with a standard deviation of only 0.0091, maintaining robust performance (97.3%) under 0.5 noise. This work advances intelligent AE monitoring for detecting damage to wind turbine blades.
Jiang et al. (Tue,) studied this question.