Sparse spatial-partition probabilistic imaging method for debonding evaluation of composite wing leading edge using ultrasonic guided wave

The composite wing leading edge (CWLE) is a critical component that determines the aerodynamic characteristics of airplanes operating in harsh environments. A large CWLE, composed of multiple material layers in a curved, sandwiched structure with varying thicknesses, is prone to debonding during both manufacturing and service. Conventional ultrasonic C-scans using bulk waves require extensive time for offline testing, making it essential to develop an efficient debonding evaluation method suitable for CWLE. This study implemented a pair of PZT arrays, functioning as excitation and reception arrays, in a section of CWLE at equal intervals to create sparse regions. A novel sparse spatial-partition probabilistic imaging method was proposed to localize debonding based on changes in ultrasonic guided waves, offering advantages of speed, baseline-free operation, and multi-feature fusion. Performance disparities among PZTs due to manufacturing and installation were addressed through comparison and compensation. The damage index of the guided wave was calculated by combining amplitude and arrival time, and the weight distribution of the sensing path, with different shape factors, was multiplied by the damage index to determine the regional probability. After determining the region with the highest probability, the probabilities on the paths between the PZT within this region and its opposing PZTs were given more weights, ultimately yielding a probability image that clearly pinpoints the debonding location. The debonding results were validated through CT scanning with a localization error of 5.02 mm. A comparison with conventional guided wave-based localization methods demonstrates that the proposed method achieves superior damage localization accuracy while maintaining equally high efficiency. The method is also validated under different measurement areas within a larger scale CWLE, showing the similar root mean square errors as that from the small scale CWLE. It suggests that the proposed method offers an efficient and accurate approach to debonding evaluation in complex material structures such as CWLE.