基于FastICA的阶次倒谱分析在齿轮箱故障诊断中的应用研究

Study on Gearbox Fault Diagnosis Using Order Cepstrum Analysis Based on FastICA

声测法作为一种非接触测量,应用于齿轮箱故障诊断时有着极大优势,但由于声信号易受到环境干扰从而影响诊断精度。将FastICA与阶次倒谱相结合,提出了一种基于FastICA的阶次倒谱方法。首先对瞬态声信号运用FastICA算法进行分离处理,分离出包含故障信息的有用源信号,然后对估计的源信号进行角域重采样,得到角域伪稳态信号,再通过倒谱分析得到基于FastICA的阶次倒谱,最终可得到故障的特征参量。应用于齿轮箱瞬态过程中的故障声信号分析,增强了信噪比,找到了故障特征,提高了齿轮箱声测故障诊断的精度。

As a non-contact measurement technique, acoustic measure has great advantage when applied in gearbox fault diagnosis, but it can be interfered by environment easily, thus affect the diagnosis accuracy. This paper combined the FastlCA with the order cepstrum analysis, proposed the order cepstrttm method based on FastlCA, Firstly, the transient acoustic signal is separated and processed by FastlCA algorithm, separate the useful source signal containing the fault information, then resampling the estimated source signal in the angular domain, obtain angular domain pseudo-stationary signals, get order cepstrum based on FastlCA through cepstrum analysis, thus obtained fault characteristic parameters. Applied in the gearbox fault acoustic signal analysis, enhanced the signal-to-noise ratio, found the fault character, and thus improved the accuracy of gearbox fault diagnosis finally.