【摘要】 针对Hilbert-Huang变换(HHT)方法出现的端点效应问题,提出了添加极值点与支持向量回归机相结合的延拓方法对其进行抑制。首先利用支持向量回归机预测信号极值点的幅值,然后用添加极值点法确定所预测极值点的位置。该改进方法解决了镜像延拓法对端点不是极值点的短数据序列处理效果不佳问题。通过仿真对比分析,用传统的镜像延拓和支持向量回归机延拓后得到的能量误差分别为0.007 4和0.023 8,而用改进方法延拓后得到的能量误差为0.003 0,表明改进方法对HHT端点效应的抑制效果要好于两种传统的方法。最后,将改进方法应用到电机转子不平衡故障特征提取上,验证了其可行性和准确性。更多还原

【Abstract】 In order to solve the endpoint effect in the Hilbert-Huang transform( HHT),an improved method is proposed by combining the adding extreme points with the support vector regression. The improved method uses the support vector regression method to predict extreme points on both ends of the original signal,and then uses the adding extreme point method to determine the position of the predicted extreme points. The improved method can solve the problem which the boundary of the short time sequence is not the extreme point by using the mirror extension method. Through comparison analysis of simulation signal,the energy error obtained by using the mirror extension and the support vector regression is 0. 0074 and 0. 0238 respectively,while the energy error is 0. 0030 by using the improved method. It indicates that the improved method can inhibit the endpoint effect of the HHT,and it is better than those two traditional methods. Finally,the improved method is used to extract the feature frequency of the motor rotor imbalance fault. The result shows that the improved method is feasible and accurate.更多还原