Abstract
Wind turbine yaw bearings play a critical role in power generation, and physical damage of these components directly incurs severe safety risks. To address the challenges of yaw bearing damage identification, a novel full vector adaptive swarm decomposition (FVASWD) method is proposed. The key influencing parameters of swarm decomposition (SWD) are adaptively optimized through the zebra optimization algorithm (ZOA). Raw orthogonal vibration signals from dual channels are processed by the optimized SWD to extract informative oscillatory components. Because directional bias is inherently present in individual sensors, a synergistic enhancement strategy based on full vector information fusion is employed to synthesize the extracted components. This process effectively enhances noise suppression under the investigated conditions, and a comprehensive full vector envelope spectrum is subsequently generated. Damage identification is performed by comparing theoretical characteristic frequencies with extracted spectral peaks. Experimental investigations demonstrate that damage characteristics can be effectively captured from complex raw signals. Quantitative results show that the proposed method achieves a fault-to-noise ratio (FNR) of approximately 26.6, demonstrating the robust feature extraction capability of proposed method under the investigated experimental conditions and its engineering applicability.