基于总体平均经验模态分解的主动噪声控制系统研究

摘要

为了提高宽窄带混合噪声的消噪效果,本文提出一种基于总体平均经验模态分解(Ensemble empirical mode decomposition, EEMD)的主动噪声控制(Active noise control, ANC)系统,利用实时EEMD 算法逐段将混合噪声分解成若干个固有模态函数(Intrinsic mode functions, IMF)分量。因为这些IMF分量的频带各不相同,所以实现了混合噪声中宽带分量和窄带分量的有效分离,独立进行ANC处理后成功解决了处理混合噪声时带来的“火花”现象,而且避免了传统混合ANC (Hybrid ANC, HANC)系统中频率失调的影响。 EEMD算法也是对混合噪声的平稳化处理过程,因此当混合噪声中出现非平稳变化时,本文提出的系统也能保持较好的系统稳定性。通过不同噪声环境下进行仿真分析,提出的ANC 系统比HANC 系统具有更好的系统稳定性和更小的稳态误差。%In order to obtain a better de-noising performance of mixture noise containing both wideband components and nar-rowband components, a new active noise control (ANC) system based on ensemble empirical mode decomposition (EEMD) is proposed in this paper. Real-time EEMD algorithm is used to decompose the mixture noise into several intrinsic mode func-tions (IMF) which have a different frequency range each other, so this decomposition can separate wideband components and nar-rowband components from the mixture noise adaptively. Each component controlled independently can not only process mix-ture noise without “firework”, but also avoid the frequency mismatch occurring in conventional hybrid ANC (HANC) sys-tem. The EEMD algorithm can smooth the mixture noise to make sure the proposed system has better stability when non-stationary phenomenon happens in mixture noise. Compared with HANC system, the proposed ANC system has better sys-tem stability and smaller steady-state error in processing differ-ent noise.