多尺度加权CEEMD-1DCNN旋转机械故障诊断

Multi-scale Weighted CEEMD-1DCNN Rotating Machinery Fault Diagnosis

旋转机械振动信号具有较强的非线性、非平稳性的特点,互补集合经验模态分解(CEEMD)克服了传统EEMD的缺陷,提供了对信号从粗到精不同尺度的刻画。针对不同尺度对故障特性描述的差异,提出一种基于多尺度加权CEEMD的一维卷积神经网络(1DCNN)故障诊断方法。利用互补集合经验模态将振动信号分解成一系列本征模态函数(IMFs),然后求取各个IMF分量的峭度值,计算各分量峭度所占权重,根据各个分量权重值对信号进行重构。将数据样本划分为训练集、验证集和测试集,将训练集输入到一维卷积神经网络中学习更新网络参数,然后用验证集进行验证得到最优诊断模型,最后利用测试集对诊断模型进行测试。通过电机轴承数据集和齿轮箱数据集两组实验进行了模型验证,诊断精度分别为99.98%和99.73%,表明所提方法能够快速准确地诊断出不同故障类型,并且具有较高的故障诊断准确率和鲁棒性。

Rotating mechanical vibration signal has strong nonlinearity,non-stationary characteristics,complementary set empirical mode decomposition(CEEMD) overcomes the shortcomings of traditional EEMD,provides the characterization of different scales of signal from coarse to fine.A fault diagnosis method based on multi-scale weighted CEEMD was proposed for the difference in fault characteristics.The vibration signal was decomposed into a series of intrinsic modal functions(IMFs) using the complementary set empirical mode,then the steepness value of each IMF component was obtained,the weight of each component steepness was calculated,and the signal was reconstructed according to the weight value of each component.The data samples were divided into training sets,validation sets and test sets.The training sets were input into the one-dimensional convolutional neural network to learn and update the network parameters.Then the validation sets were used to verify the optimal diagnostic model.Finally the diagnostic models were tested by the test set.The model verification was carried out by two sets of experiments,the motor bearing data set and the gearbox data set,and the diagnostic accuracy was 99.98% and 99.73% respectively.The results show that the proposed method can quickly and accurately diagnose different fault types,and has a high fault diagnosis accuracy and robustness.