【摘要】 针对轴承故障诊断中，特征提取环节严重依赖人工经验及专家知识的问题，提出了一种基于格拉姆角场（Gramian angle field,GAF）变换和自适应深度网络的轴承故障诊断方法。首先，通过经验模态分解方法对采集信号进行分析，通过马氏距离度量方法有效地确定本征模函数（Intrinsic mode functions,IMFs），将染噪信号与原始信号的相似模态分量进行挑选，以提高染噪信号信噪比，剔除不同类别信号的相似模态分量，突出信号特征；然后，利用选定的IMFs将信号重构，并基于GAF变换将重构信号可视化；最后，利用深度自适应网络进行特征学习和状态识别。结果表明，所提方法的准确率达到94.97%，优于常见的振动信号故障诊断方法；且所提方法对于噪声也能很好地抑制，具有较好的鲁棒性，为轴承的智能化和精确化诊断提供了合理思路。更多还原

【Abstract】 Aiming at the problem that feature extraction in bearing fault diagnosis needs to rely heavily on manual experience and expert knowledge,a bearing fault diagnosis method based on Gramian angle field(GAF)transformation and adaptive depth network is proposed. Firstly,the collected signals are analyzed by empirical mode decomposition method,and the Intrinsic mode functions(IMFs) are effectively determined by Mahalanobis distance measurement method. It can not only select the similar modal components of the noisy signal and the original signal to improve the signal-to-noise ratio of the noisy signal,but also eliminate the similar modal components of different types of signals to highlight the signal characteristics;then,the signal is reconstructed by using the selected IMFs,and the reconstructed signal is visualized based on GAF transform;finally,the depth adaptive network is used for feature learning and state recognition. The results show that the accuracy of the proposed method is 94.97%,which is better than the common vibration signal fault diagnosis methods,and the proposed method can also suppress the noise and has good robustness,which provides a reasonable idea for the intelligent and accurate diagnosis of bearings.更多还原