【摘要】 为了提高航空发动机旋转机械故障信号特征提取效果与诊断准确率，提出了一种集合经验模态分解（EEMD）融合功率谱熵的故障诊断方法。该方法采用EEMD对原始信号进行分解，并利用功率谱熵定量分析了各阶本征模态函数（IMF）的信息量，并对部分IMF自适应降噪处理。重构所有IMF与余项，并输入至卷积神经网络（CNN）进行训练与故障分类。分别利用理想信号与航空发动机旋转机械故障模拟平台的实测信号，验证了所提出的信号处理方法与故障诊断方法的有效性与优势。结果表明：相较于传统信号处理与故障诊断方法，该方法处理信号后的信噪比（SNR）提高25%以上，均方误差（MSE）减小40%以上，故障诊断准确率提高10%以上，更有利于工程中的旋转机械故障定位与诊断。更多还原

【Abstract】 In order to improve the feature extraction effect and the diagnostic accuracy of aeroengine rotating machinery fault signal,a fault diagnosis method based on EEMD and power spectrum entropy was proposed. Firstly,the original signal was decomposed by EEMD,the power spectrum entropy was used to quantitatively analyze the information of each order of the IMFs,and some IMFs were adaptively denoised. Secondly,all IMFs and residual items were reconstructed and input into the CNN for training and fault classification.Finally,the effectiveness and superiority of the proposed signal processing method and the fault diagnosis method were verified by using the ideal signal and the measured signals from the aeroengine rotating machinery fault simulation platform, respectively. The results show that compared with the traditional signal processing and fault diagnosis methods,the SNR of the signal processed by the proposed method is increased by more than 25%,the MSE of the signal is reduced by more than 40%,and the fault diagnostic accuracy is increased by more than 10%,which is more conducive to the fault locating and diagnosis of rotating machinery in engineering.更多还原