基于SSAE的输入级弧齿锥齿轮自动特征提取及故障诊断

Automatic Feature Extraction and Fault Diagnosis of Input-stage Spiral Bevel Gear Based on SSAE

航空用输入级弧齿锥齿轮常处于高速重载工况,获取的振动信号具有强非线性、非平稳特点,造成故障特征难提取。对此,运用堆栈稀疏自动编码器故障自动特征提取方法结合分类器对输入级弧齿锥齿轮故障诊断进行了研究。搭建输入级弧齿锥齿轮故障诊断模拟试验台,分别进行正常齿轮和故障齿轮运行的测试试验。将多个SAE层层堆叠形成SSAE,对弧齿锥齿轮故障特征层层提取,用多分类器完成故障诊断。结果表明:该方法输入级弧齿锥齿轮故障识别结果的有效率可达100%,为弧齿锥齿轮的故障分析提供了一种有效途径。

As aviation input-stage spiral bevel gears are often in high-speed and heavy-load conditions, and the acquired vibration signals have strong nonlinear and non-stationary characteristics, it is difficult to extract fault characteristics. To overcome the difficulty, the fault diagnosis of input-stage spiral bevel gears is studied by the automatic feature extraction method of stack sparse autoencoder and the classifier. The input-level spiral bevel gear fault diagnosis simulation test bench was built, the normal and faulty gear functions were tested, multiple SAEs were stacked to form SSAE, spiral bevel gear fault characteristics were extracted layer by layer, and themulti-classifier was applied to complete fault diagnosis. The results show that the effective rate of the input-stage spiral bevel gear fault identification results of this method can reach 100%, which provides effective means for the failure analysis of spiral bevel gears.