Abstract: In order to solve the difficulty of bearing weak compound fault feature extraction under strong background noise, a bearing compound fault diagnosis method based on AVMD and WDK for wind turbine gearbox is proposed in this paper. Firstly, an adaptive learning particle swarm optimization （ALPSO） algorithm is introduced, and the average envelope entropy is adopted as the fitness function of ALPSO to search for the optimal influence parameters of the variational mode decomposition, thus the adaptive variational mode decomposition （AVMD） is constructed. Secondly, the Wasserstein distance kurtosis（WDK） index is proposed to screen the effective modal components combining the advantages of Wasserstein distance and kurtosis, and the selected effective modal components are reconstructed. Thirdly, the reconstructed signal is analyzed through envelope spectrum analysis to realize the bearing compound fault diagnosis. Finally, the AVMD-WDK method is applied to the bearing fault diagnosis for a 2 MW wind turbine gearbox in a wind field. The experimental results show that the proposed method can effectively extract the weak fault features of bearings and realize the bearing compound fault diagnosis accurately.

Key words: wind turbine, compound fault, gearbox, adaptive variational mode decomposition, Wasserstein distance kurtosis

摘要： 针对强背景噪声下轴承微弱复合故障特征提取困难的问题,提出一种基于自适应变分模态分解（AVMD）和优化的Wasserstein距离指标（WDK）的风电齿轮箱轴承复合故障诊断方法。首先,引入自适应学习粒子群优化算法（ALPSO）,以平均包络熵作为ALPSO的适应度函数来搜索变分模态分解的最佳影响参数,从而构造AVMD;其次,结合Wasserstein距离（WD）和峭度优点,提出WDK指标筛选有效模态分量,并对筛选的有效模态分量进行重构;然后,通过对重构信号进行包络谱分析实现轴承复合故障的诊断;最后,将所提AVMD-WDK方法应用于某风场2 MW风电齿轮箱轴承振动信号的故障诊断。结果表明,该方法能有效提取轴承的微弱故障特征,实现轴承复合故障的精确诊断。

关键词: 风电机组, 复合故障, 齿轮箱, 自适应变分模态分解, 优化的Wasserstein距离指标（WDK）