ICA在航空发动机振动信号盲源分离中的应用

研究了基于独立分量分析(independent component analysis,简称ICA)的发动机振动信号盲源分离技术,旨在将发动机振动信号按照不同的激振源进行分离。首先阐述了基于最大信噪比的盲源分离算法原理,通过对仿真信号进行分离,判断了分离输出信号与仿真信号的一致性,验证了该算法的可行性;然后将该算法与FFT分离法相结合,应用于某型双转子航空发动机高、低压转子实测振动信号盲源分离中,取得了很好的分离效果,表明应用ICA技术建立的基于最大信噪比的盲源分离算法具有迭代次数少、计算复杂度低、效果好及稳定等优点。

Blind Source Separation for Aero-engines Vibration Signal by Independent Component Analysis

Abstract The measured signals from the casing of an aero-engine are the mixture of the vibration caused by the low pressure rotor, the high pressure rotor and the other rotating parts. In order to separate the vibration reduced by every different vibration source, this paper investigated aero-engine vibration signal blind source separation method based on the independent component analysis. Firstly, a blind source separation algorithm based on maximum signal to noise ratio was expounded in this paper, the algorithm was adopted to separate initial signals and the output was consistent well with the source signal. It was validated that the algorithm was of correctness and feasibility. Secondly, combination of the above algorithm with FFT separation method was used in the blind source separation of measured vibration signals. It was proved that the method was effective to separate the high-pressure rotor vibration and the low-pressure rotor vibration of a twin-spool aero-engine. It has some advantages, such as less iteration times, simple calculation process and a high stability. This paper presents a feasible and effective signal processing method of fault diagnosis and vibration monitoring for twin-spool aero-engines.