Abstract:

For the end effects of EMD, an improved extension method was proposed based on boundary local features. By calculating the mathcing errors between the boundary wave and any internal signal wave, the regularity strength of signal was determined. In the case that the inherent regularity was strong, the internal wave form matching extension was conducted to maintain the inherent tendency to the greatest extent. While the signal was weak regularity and the signal edge was changed abnormally, the local information at the edges was considered only to conduct the extreme point extension. Thus the extreme points were used to fit envelope curve to realize extension. This method ensures that the decomposition of valid data is correct, and improves the decomposition accuracy. Simulation results show that the method can suppress end effect of EMD method.

Key words: empirical mode decomposition(EMD), end effect, characteristic scale extension, torsional vibration signal of rolling mill

摘要：

针对经验模态分解(empirical mode decomposition,EMD)中存在的端点效应问题，提出一种边界局部特征尺度延拓的EMD改进方法。通过计算信号边缘处与信号内部任何一段子波的匹配误差，确定信号的规律性强弱。在信号内在规律性较强的情况下，进行内在的匹配波形延拓，最大限度地维护信号的内在趋势；在信号规律性较弱的情况下，特别是信号边缘发生异常变化的情况下，只考虑边缘处的局部信息，根据边缘局部极值点的特征进行延拓，对延拓的极值点序列进行包络拟合，估计出均值曲线。该方法保证了信号有效数据的正确分解,提高了信号的分解精度，实现了EMD算法的改进。仿真实验验证了该方法能较好地抑制EMD端点效应。

关键词: EMD, 端点效应, 特征尺度延拓, 轧机扭振信号