航空发动机转静径向碰摩位置的智能识别技术研究

获取航空发动机转静碰摩位置对于诊断发动机碰摩故障和改进设计具有重要意义，基于航空发动机转子实验器的机匣振动加速度信号，研究了一种基于拉普拉斯特征映射（Laplacian Eigenmaps，LE）结合球结构支持向量机的径向碰摩位置智能识别方法。首先，采用拉普拉斯特征映射算法提取碰摩样本的特征信息，并用网格搜索法优化拉普拉斯特征映射算法的相关参数；然后，将特征样本作为输入到球结构支持向量机分类器，识别不同位置的碰摩样本。最后，利用实测的碰摩数据对此方法进行了验证，并与主成分分析法(PCA)所得特征样本的分类结果进行了比较，结果表明了方法的实用性和有效性。

Study on Intelligent Recognition of the Aero-engine Rotor-Stator Radial Rubbing location

It is very important to acquire the aero-engine rotor-stator radial rubbing location, in order to diagnosis the rubbing fault and improve the design. Based on the casing vibration acceleration signal of the aero-engine rotor experimental rig, a method for radial rubbing location identification based on Laplacian Eigenmaps (LE) and sphere support vector machine was investigated in this paper. First, the Laplacian Eigenmaps was used to extract the rubbing samples’ features, which parameters were optimized by grid search method; and then, the characteristics of the samples were input to the sphere support vector machine to identify different locations rubbing samples. Besides, with the actual rubbing data the method was confirmed and compared with Principal component analysis (PCA), and the results show the practicability and effectiveness of the method.