基于EMD和多特征组合的液压信号辨识方法

发布时间：2018-05-22 20:25

  本文选题：液压信号 + EMD　； 参考：《液压与气动》2015年11期

【摘要】：液压信号具有非平稳性、非线性、特征信息相近时难以正确辨识的特点。针对该特点提出了一种经验模态分解(EMD)和多特征组合的信号辨识方法。该方法将信号自适应分解为若干个固有模态函数(IMF);提取各IMF分量的能量、裕度、峰度、波动系数等特征参数,规范化后组合形成全局特征向量,并输入支持向量机(SVM)中学习和辨识。通过对液压主管压力信号处理表明:该方法能有效辨识特征信息相近的压力信号,在小样本下仍然具有较好的辨识率。
[Abstract]:Hydraulic signals are non-stationary, nonlinear and difficult to identify correctly when the characteristic information is close. A signal identification method based on empirical mode decomposition (EMD) and multi-feature combination is proposed. The method decomposes the signal adaptively into several inherent mode functions, extracts the energy, margin, kurtosis and fluctuation coefficient of each IMF component, and forms a global eigenvector after normalization. And input support vector machine (SVM) learning and identification. The pressure signal processing of hydraulic supervisor shows that the method can effectively identify the pressure signal with similar characteristic information, and it still has a good identification rate under small samples.
【作者单位】： 武汉科技大学信息科学与工程学院;
【基金】：国家自然科学基金(61174106)
【分类号】：TG333.1
，

本文编号：1923489