Wind turbine blades are critical components in wind energy generation, and blade health management is a challenging issue for the operation and maintenance of wind turbines. In this paper, an adaptive method is developed to identify blade damages based on the microphone array and compressive beamforming, and global and remote health assessment can be accomplished. In this method, the generalized minimax-concave penalty function is employed to enhance sparse recovery capacities, and step-sizes in computation processes are adjusted adaptively to adapt to variational conditions. Besides, potential damage locations are extracted in coarse acoustic maps to improve convergence rates. Numerical simulations show that high spatial resolutions can be achieved by the proposed method, and the computation time for solving acoustic inverse problems is less than using existing algorithms, especially with low-frequency sources. Moreover, experiments are conducted with a small-scale wind turbine. Results demonstrate that several damages in operating blades can be precisely recognized with high efficiencies, and the deterioration of acoustic maps induced by improper step-sizes can be avoided. The proposed method provides a promising way for in-situ health monitoring of wind turbine blades.

风力涡轮机叶片是风力发电的关键部件，叶片健康管理是风力涡轮机运行和维护的一个具有挑战性的问题。在本文中，开发了一种基于麦克风阵列和压缩波束形成的自适应方法来识别叶片损坏，并且可以完成全局和远程健康评估。该方法采用广义极小极大凹惩罚函数来增强稀疏恢复能力，并自适应调整计算过程中的步长以适应变分条件。此外，在粗略的声学图中提取潜在的损坏位置以提高收敛速度。数值模拟表明，所提出的方法可以实现高空间分辨率，并且解决声学逆问题的计算时间少于使用现有算法，尤其是对于低频源。此外，实验是用小型风力涡轮机进行的。结果表明，可以高效准确地识别运行中的叶片中的几种损坏，并且可以避免因步长不当引起的声学图的恶化。所提出的方法为风力涡轮机叶片的原位健康监测提供了一种有前途的方法。