基于小波包分析和支持向量机的转静碰摩部位识别

利用航空发动机转子实验器模拟不同径向碰摩部位下的碰摩故障,提出基于小波包分析的支持向量机转静碰摩部位识别方法.首先将从机匣测得的加速度信号进行小波包分解,提取其归一化能量特征,接下来将得到的归一化能量特征输入至支持向量机中,用以识别不同的碰摩部位.利用航空发动机转子实验器模拟大量不同碰摩程度和不同碰摩部位的样本,利用支持向量机进行训练和测试.结果表明小波包能量特征与支持向量机相结合可以有效地判别转静碰摩部位,且仅需1个传感器即可达到98％的识别率.

Rotor-stator rubbing positions identification of aero-engine based on wavelet packet analysis and support vector machine

Rubbing faults of different radial rubbing positions by using the rotor experiment rig of aero-engine are simulated.An identification method based on wavelet packet analysis and support vector machine(SVM) was proposed.Firstly,the acceleration signals on the casing were collected.Secondly,the signals were decomposed by wavelet packet analysis,and the normalized energy features were extracted.Finally,the normalized energy characteristics were input into support vector machine to identify the different rubbing positions.By using an aero-engine rotor experiment rig,a large number of samples including different rubbing positions and different rubbing degrees were simulated, and the support vector machine was trained and tested by these samples.The results show that the new method combining the wavelet packet energy features and support vector machine can effectively identify the rotor-stator rubbing positions of aero-engine;in addition,only one sensor is required to reach the recognition rate of 98%.