主频占优镶边滤波器分解信号为本征模态函数的方法

A method on preferential dominant frequency and mount edge filter decomposing signal into intrinsic mode functions

针对本征模态函数分解方法FFDSI存在的问题,首先将信号变换到频率域,采用以主频率为始点逐渐向两边扩大频率通过带宽度的方法,寻求一种最宽带通镶边滤波器,使信号经此滤波器滤波后得到的信号为本征模态函数。然后,从原信号减去此模态函数并重复这一过程,便可实现信号的本征模态函数分解。新方法不仅可以有效削弱吉布斯效应,较好地反映信号的瞬时特性,尽可能地降低拆分"拍"信号的机率,而且在分解过程中还同时得到了本征模态函数的解析信号,这为以后计算Hilbert谱提供了很大便利。文中还对风浪信号进行了分解,得到了5个有意义的主要本征模态函数。

For the problems existing in decomposition method of FFDSI, firstly the signal is transformed from time to frequency domain, and then based upon a method, through which the main frequency is taken as the initial point expanding pass band to both sides gradually, the widest pass band mount edge filter is found and then the filtered signal is turn to be the intrinsic mode function. Next, the mode function is subtracted from the raw signal and the above process is repeated, thus, the raw signal is decomposed into a series of intrinsic mode functions. The new method can not only weak the Gibbsr phenomenon effectively, reflect the instantaneous character better, reduce the chance to split ＂beat＂ signal as far as possible, but also get the analytical signals of the intrinsic mode functions simultaneously, which will provide a great convenience in later calculation of Hilbert spectrum. In the paper, the new method is used in the wind wave signal decomposition and the five main meaningful intrinsic mode functions are obtained.