独立分量分析方法在经验模式分解中的应用

Application of independent component analysis in empirical mode decomposition

若信号间的能量和频率比例过大,经验模式分解不能分解出正确的单一模式分量。针对这种状况提出一种经验模式分解与独立分量相结合的信号分析方法。该方法能分离出IMF分量的固有特性,消除EMD分解过后各IMF之间信息混淆问题,恢复各个单分量所丢失的信息特性,改善了经验模式分解能力不足所带来局限性,保障经验模式分解的有效性。通过仿真信号和实际工程信号研究,验证了该方法的可行性。表明该方法对信号分解和故障诊断具有很好的前景。

If ratio of signal's energy and frequency is much larger,the method of empirical node decomposition(EMD) will not work normally.A the new integrated method of EMD and independent component analysis(ICA) was presented here.It can seprate an overlapped information,decompose accurately intrinsic mode functions(IMFs),recover characteristic of an IMF,and perfect deficiencies of EMD's capability.This method can protect the validity of EMD.The feasibility of this method was demonstrated through application of simulated signals and practical signals.The results showed that this method is useful in signal processing and fault diagnosis of mechanical equipments.