利用ICA的局域波分解及其在机械故障诊断中应用

Local wave decomposition using ICA and its application to fault diagnosis in machinery

局域波分解算法的自身局限使得分解分量之间的正交性不能得到保证,分解分量存在信息冗余,从而影响了分解质量.针对这个问题,将独立分量分析(ICA)引入局域波分解过程中,以消除分解分量的信息冗余,使得局域波分解得到的基本模式分量之间相互正交独立,改善局域波分解算法的局限性,准确地进行局域波分解.应用该方法对转子早期碰摩振动信号进行了分析,并对分解得到的高频基本模式分量进行Hilbert包络解调分析,准确提取了转子碰摩故障特征.试验数据的分析结果证明了方法的准确性和有效性.

Due to its own implicit difficulties, the local wave decomposition （LWD） algorithm cannot guarantee the completeness and orthogonality of the components, and may produce redundant components and affect the accuracy of decomposition. In order to reduce the components redundancy in LWD, independent component analysis （ICA） is introduced into LWD, so that the components are orthogonal to each other and the components redundancy can be reduced. Finally, a much better decomposition performance can be obtained. It is employed to analyze the vibration signal of rotor system with early rub-impact for extracting the weak impulsive feature. The signal is decomposed into intrinsic mode functions accurately, then the high-frequency components are analyzed by Hilbert envelope demodulation. The modulating fault feature of vibration signal of a rotor system with rub- impact fault can be extracted exactly. Analytical results of experimental data show that the proposed method is accurate and efficient.