基于分形关联维的汽轮机转子的振动故障诊断

Fault Diagnosis of Steam-Turbine Rotor Vibration Based on Fractal Correlation Dimension

采用Bently实验台模拟不同转速下碰摩、松动、不对中、不平衡几种常见汽轮机转子振动故障,进行实验研究.通过相空间重构理论,对满足随机分形统计自相似性的振动信号序列进行相空间重构,计算其分形关联维数,从而再现动力学特性.实验分析结果表明:碰摩故障时的分形关联维数最大,松动和不对中时的分形关联维数次之,不平衡时计算所得的分形关联维数最小;不同故障类型下的分形关联维数增量不同.因此,分形关联维数可以作为一种振动故障征兆加以提取.

This paper experimentally investigates the vibration faults of the steam-turbine rotor, such as the rubbing, the loosening, the uncountershaft and the mass unbalance, using the Bently experiment table. According to the theory of Phase Space Reconstruction, the vibration signal serial, a kind of random fractal that meets the statistical self-similarity, is reconstructed. By the calculation of fractal dimension, the dynamic characteristics of the system are then reproduced. The results show that the fractal correlation dimension of rubbing fault is the largest, while those of the loosening fault and the uncountershaft fault are a little lower, and that of the mass unbalance fault is the smallest. It is also shown that different types of faults result in different fractal correlation dimension increments. Therefore, the fractal correlation dimension can be abstracted as a sign of vibration fault.