基于最小熵解卷积的齿轮箱早期故障诊断
Incipient Fault Diagnosis of the Gear box Based on the Minimum Entropy Deconvolution

齿轮箱发生早期故障时，其振动信号一般很微弱，且隐含的能反应出齿轮箱运转状态的冲击成分常被淹没在强烈的噪声中，直接做频谱分析或包络谱分析，很难提取其故障特征。论文将最小解卷积方法应用于炼胶机的齿轮箱故障诊断。首先利用该方法对齿轮箱振动信号进行解卷积滤波处理，然后对滤波后的信号进行包络解调分析，最后提取出了该齿轮箱轴5上齿轮8（z8=28）齿根轻微裂纹的故障特征，实现了该齿轮箱的早期诊断。应用实例验证了最小熵解卷积方法的有效性和优点。
Comparing with the strong background noise, the gear's vibration signal is usually very weak when occurring the incipient fault of gear box, and with concealed impulse component reflecting the operating state of the gearbox. The vibration feature is very difficult to extract through frequency spectrum or envelope spectrum analysis. The minimum entropy deconvolution (MED) was adopted to carry out the fault diagnosis of gear box for rubber refiner. The vibration signal of gearbox was firstly filtered with the MED method, then the filtered signal was demodulated with the envelope spectrum, and the fault feature of root slight crack of gear 8(z8=28) on the shaft five was successfully extracted. Therefore, the incipient fault diagnosis of the gear box was realized. The application results verify the effectiveness and advantage of MED method