针对经验小波变换（Empirical Wavelet Transform，EWT）对强噪声环境下风力机齿轮箱轴承轻微故障特征提取不足的问题，利用滑移窗口提取子带的连续平均谱负熵（Continuous Average Spectral Negentropy，CASN）对EWT算法进行改进。通过CASN-EWT方法分解风力机齿轮箱轴承故障信号，采用峭度准则对所得分量进行筛选并重构，再开展包络分析，准确提取出故障特征。结果表明：CASN-EWT方法在保留EWT算法自适应性和有效避免模态混叠效应与端点效应优点的同时，能够极大提高EWT分解算法对噪声影响的鲁棒性，有利于准确提取故障特征频率，实现故障精确识别。

Aiming at the problem of weak fault features of wind turbine gearbox bearings insufficiently extracted by empirical wavelet transform (EWT) under the background of strong noise,a sliding window is used to extract the continuous average spectral negentropy (CASN) of the subbands to improve the EWT algorithm.The CASNEWT method is used to decompose the fault signals of the wind turbine gearbox bearing,and then the acquired components are filtered and reconstructed by the kurtosis criterion,and the envelope analysis is carried out to accurately extract the fault characteristics.The results show that the CASNEWT method retains the adaptivity of the EWT algorithm,which can effectively avoid the modal aliasing effect and the end effect,and meanwhile greatly improve the robustness of the EWT decomposition algorithm to noise,contributing to accurately extract the frequency of the fault characteristic to realize accurate fault identification.

TK83

国家自然科学基金（51976131，51676131）；上海市“科技创新行动计划”地方院校能力建设项目（19060502200）