摘要
为进一步提高航空发动机振动状态监测的有效性和故障诊断的准确性,将机匣截面振动信号的各谐波轴心轨迹椭圆长短轴乘积看成广义时间序列。基于该序列能够全面反映发动机转子系统各谐波能量分布的客观事实,利用其构造矩阵并提取奇异值向量。借助于该向量构造特征值,通过比较特征值向量实现对发动机不同振动状态的识别。对实测振动信号的分析表明:在同一振动状态下,各数据椭圆长短轴乘积相对奇异值强度具有相同的变化趋势和良好的稳定性;在不同振动状态下,椭圆长短轴乘积相对奇异值强度变化趋势不尽相同;通过椭圆长短轴乘积奇异值相对距离熵能够较好地识别发动机各振动状态。
In order to improve the accuracy of vibration monitoring and fault diagnosis of aeroengine, a new vibration state recognition method was brought out. Based on the fact that the different harmonic energy of aeroengine rotor system, the method can be completely presented by the product of major axis and minor axis of shaft centerline orbit ellipse. Different harmonic products were treated as general time series. The SVD theorem was used to the array which was formed by the time series. Different vibration states of aeroengine can be identified by characteristic vector which was derived from eigenvalues of the array. The method was used to analysis the real testing data.The result shows that the relative intensity of product singular value has the same varied tendency and better stability in the same vibration state, while the tendency is distinct in different states and different vibration states can be recognized by the relative distance entropies.
出处
《航空发动机》
2016年第3期38-42,共5页
Aeroengine
基金
航空动力基础研究项目资助
关键词
振动能量积
奇异值分解
状态识别
振动信号
航空发动机
vibration energy product
singular value decomposition
state recognition
vibration signal
aeroengine