利用SG滤波算法将能量贡献率低于0.1%且大于0.05%的IMF进行去噪处理.实验结果表明,CEEMDAN-SG与经验模态分解(EMD),改进的集合经验模态分解(EEMD),CEEMDAN以及CEEMDAN-小波阈值去噪所比较,信噪比分别提高了0.85,0.71,3.09,0.25 dB,且均方误差最小.CEEMDAN-SG与CEEMDAN-小波阈值去噪在去除噪声效果较理想,且CEEMDAN-SG在...
autocorrelation coefficient and Person correlation coefficient were used to divide the IMF components.A novel noise reduction scheme is proposed: the original SOH is decomposed using CEEMDAN, and reconstructs the dominant noise components after SG filtering to obtain the SOH sequence after noise reduction...
基于水电机组振动信号存在的非平稳和非线性,提出一种结合二次分解和简单循环单元(SRU)与注意力结合(Attention)的方法.首先采用ICEEMDAN和VMD方法对振动信号进行二次分解,将信号分解成本征模态函数(IMF);随后,引入基于SRU的深度学习模型,将提取得到的IMF信号输入到SRU中进行学习得到水电机组故障识别器.为了进一步提升模型...