基于NABEMD的飞行参数噪声能量估计与消除

Noise Energy Estimation and Denoising of Flight Data Based on Noise Assisted Bivariate Empirical Mode Decomposition

飞参系统(Flight Data Recorder,简称FDR)记录的飞行参数采样率低、噪声与野值混杂且呈非线性非平稳性的特点使得传统噪声处理方法难以适用,因而提出一种基于噪声辅助复数据经验模态分解(Noise Assisted Bivariate Empirical Mode Decomposition,简称NABEMD)的噪声能量估计与消除方法,用于飞行参数的降噪问题。该方法首先利用飞行参数和高斯白噪声构造复数据并进行BEMD分解,然后根据虚部各层内禀模态函数(Intrinsic Mode Function,简称IMF)的能量来估计实部IMF包含的噪声能量,最后根据噪声能量估计值对IMF进行分层处理得到降噪后的信号。仿真结果表明,本文方法相对于现有方法具有一定优势,可以进一步提高飞行参数的降噪精度。

To solve the problems that traditional denoising methods are not suitable for processing the flight data recorder parameters which are of low sampling frequency,noise mixes with abnormity,nonlinear and nonstationary,a denoising method of flight data recorder parameters based on noise assisted bivariate empirical mode decomposition is proposed. Firstly,a complex signal constructed by given parameter and gaussian white noise is decomposed by bivariate empirical mode decomposition; secondly,according to the energy of each intrinsic mode function in imaginary part,the noise energy of each intrinsic mode function in real part is estimated; finally,the denoised parameter is obtained by dealing with each intrinsic mode function separately. Simulation results show that our method has more performance advantages than existing methods,and can improve the precision of flight data recorder parameters denoising.