中文摘要：

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。

英文摘要：

Hydraulic signal is non-stationary and non-linear and hard to identify when has similar feature information. For the feature, a signal identification method based on empirical mode decomposition （EMD） and multi-feature combination is proposed . This method decompose signals into several intrinsic mode functions （IMFs） adaptively. The IMFs energy , margin, fluctuation coefficient and other feature parameters are extracted and normalization to form global feature vector. Then the global feature vector is put into support vector machine （SVM） for learning and recognition. By processing of head pressure signals, the result shows that the method can effectively identify pressure signal of similar characteristic information, and still has good recognition rate under small sample.