论文部分内容阅读
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.
Aeroengine rotor experimental apparatus is used to simulate rub-impact faults under different radial rubbing sites, and a new method based on wavelet packet analysis is proposed to identify static rubbing sites of support vector machine.At first, the acceleration signal measured from the casing is decomposed by wavelet packet , The normalized energy features are extracted, and then the normalized energy features are input into the support vector machine to identify different rubbing sites.Using aero-engine rotor experimental apparatus to simulate a large number of different rubbing degrees and different touch The results show that the combination of wavelet packet energy feature and support vector machine can effectively distinguish the static friction rubbing site and only need one sensor to achieve the recognition rate of 98% .