具有独立分量的经验模态分解算法研究

Improved algorithm for empirical mode decomposition with independent elements

在经验模态分解算法中用极值包络平均近似局部平均,不能保证分解分量之间的正交性,固有模态分量存在冗余.这种情况对信号成份分析尤为不利,冗余部分的物理意义无法解释,或可能作出错误的解释.将独立分量分析方法引入经验模态分解算法中,利用其良好的分解独立特性,使模态分量不仅正交而且相互独立,消除了冗余.仿真试验表明,改进算法的模态分量彼此独立,特别对于混有突变信号的周期信号,在得到周期分量的同时,也得到突变分量,说明了改进算法比原算法优越,且具有较好的工程应用前景.

To solve the problem that the elements may be redundancies in the algorithm of empirical mode decomposition, the theory of independent component analysis was introduced into the empirical mode decomposition to improve the algorithm. Thus the elements are orthogonal to each other and the decomposition redundance can be cut down. Then the signal components can be expressed accurately. Simulation results show that the improved algorithm is better than common algorithms of empirical mode decomposition. The intrinsic mode function is independent to each other and the redundancies are eliminated. So the component, such as pulse or step in periodic signal, can be expressed by a single decomposition element. It is indicated that the improved algorithm has good foreground in many project fields.