摘要
针对实时变速工况下滚动轴承故障信号特征提取与在线识别难题,提出一种创新性的轴承故障诊断方法。该方法融合频带分割(Frequency Band Spliting,FBS)与经遗传算法优化的BP神经网络(Genetic Algorithm-Backpropagation Neural Network,GA-BP)。首先,对原始信号进行小波包分解得到小波包能量谱,然后提取振动信号和转速信号均值和方差指标共同构建特征参数集,为进一步降低数据采集成本,对特征参数集进行降采样处理。最后利用GA-BP的最佳隐含层自适应寻优系统,实现对故障特征的精确识别。实验和应用案例表明,其对于美国凯斯西储大学轴承数据集故障诊断准确率达100%,对于加拿大渥太华大学轴承数据集故障诊断准确率达99.4%,充分证明所提方法的经济性、有效性和良好的鲁棒特性。
Aiming at the difficulty of feature extraction and online identification of rolling bearing fault signals under real-time variable-speed conditions,an innovative bearing fault diagnosis method is proposed.The method combines Frequency Band Spliting(FBS)and Genetic Algorithm-Backpropagation Neural Network(GA-BP).Firstly,wavelet packet energy spectrum is obtained by wavelet packet decomposition of the original signal.And then,the vibration signal and rotational speed signal mean and variance index are extracted to construct the characteristic parameter set,which is downsampled to further reduce the cost of data collection.Finally,the best hidden layer adaptive optimization system of GABP is used to realize the accurate recognition of fault features.The experiments and application cases show that the fault diagnosis accuracy of the bearing dataset of Case Western Reserve University in the United States can be up to 100%,and that of the bearing dataset of the University of Ottawa in Canada can be up to 99.4%,which fully proves the economic validity and good robustness characteristics of the proposed method.
作者
王思思
蒋淑霞
陈晓飞
吴杰
黄成祥
WANG Sisi;JIANG Shuxia;CHEN Xiaofei;WU Jie;HUANG Chengxiang(College of Mechanical and Intelligent Manufacturing,Central South University of Forestry Science and Technology,Changsha 410004,China)
出处
《噪声与振动控制》
北大核心
2025年第5期138-145,共8页
Noise and Vibration Control
关键词
故障诊断
频带分割
降采样处理
实时变速工况
小波包能量谱
fault diagnosis
FBS
downsampling processing
real-time variable speed conditions
wavelet packet energy spectrum
