基于改进经验模式分解算法的实时滤波新方法

由于电液伺服系统本身的特点使某些干扰信号频段与正常信号的频段很靠近，普通数字滤波器很难起到较好的作用。为了能够精确过滤这些干扰信号，改进了Hilbert-Huang变换中所提出的经验模式分解(EMD)算法，扩展了极值点的定义，给出了插值判断条件，新增了分解结束判断条件。改进后的EMD算法对电液伺服系统信号有更好的包络效果、更佳的分解效果和更快的分解速度，可以满足实时滤波要求。将基于改进EMD算法的实时滤波器应用于实际的电液伺服控制系统后，同普通数字滤波器和原EMD算法进行对比，其结果验证了该滤波器的滤波效果优于普通数字滤波器，计算效率得到提高。

A New On-line Filtering Method Based on Improved EMD Algorithm

The normal digital filter cannot achieve a good filtering result since the characteristics of electro-hydraulic servo system makes the frequency of noise signal close to the frequency of useful signal. The empirical mode decomposition algorithm is improved to remove these noise signals exactly from data. The definition of extreme points is extended, a judging formula of envelope interpolation is given, and a new decomposition end judgment condition is added. The improved EMD method has better effect and decomposition efficiency on the signal from electro hydraulic servo system. It can meet the real-time filtering requirements. This new filter based on improved EMD method is used in electro-hydraulic servo system, and the normal digital filter is compared with the original EMD method. The test results show that this new filter has a good application effect and less consuming time.