分形维数约束ICA-RLS的遥测振动信号降噪方法

Telemetry Vibration Signal Denoising Method Based on Fractal Dimension Constraint ICA-RLS

针对噪声会对遥测振动信号时频分析产生影响的问题,提出一种分形维数约束独立分量分析ICA-递归最小均方RLS的降噪方法,首先对振动信号进行相空间重构,采用基于峭度最大化的牛顿迭代ICA算法对信号进行盲源分离;然后计算各源信号的分形维数,根据分形维数自动判别含噪源信号,对含噪源信号进行RLS自适应滤波;最后对滤波后的信号进行重构,得到降噪后的振动信号。实测信号处理结果验证了算法的有效性。

Aiming at the problem that the noise will affect the time-frequency analysis of telemetry vibration signal,a noise denoising method based on fractal dimension constraint Independent Component Analysis(ICA)-Recursive least Mean Square(RLS)was proposed.First,the phase space reconstruction was used for vibration signal,and the kurtosis maximization ICA algorithm based on Newton iterative was used to blind separate the signal.Then the fractal dimension of each source signal was calculated,the noises was judged automatically according to the fractal dimension,and denoised by RLS adaptive filtering.Finally,the filtered signal was reconstructed,and the measured signal processing results verified the effectiveness of the algorithm.