基于支持向量机和窗函数的DEMD端点效应抑制方法

End Effects Restraining of DEMD Based on Support Vector Machine and Window Function

在微分经验模式分解过程中,由于信号端点处极值点不确定,导致在样条曲线构造信号上下包络线的过程中产生端点效应,影响微分经验模式分解的质量。针对此问题,提出一种支持向量机延拓和窗函数相结合的方法来解决微分经验模式分解中的端点效应,通过采用支持向量机对信号两端进行数据延拓,再对延拓后的信号加特殊窗函数处理,减小延拓误差。通过仿真分析和滚动轴承故障诊断实例分析表明,该方法能较好地抑制微分经验模式分解的端点效应,提高信号分解的精度,得到准确的分析结果。

In the process of differential-based empirical mode decomposition （DEMD）, as a result of the uncertainty of the signal at both ends of the border, the end effects exist in the course of getting envelop of data, it will affect the quality of differential-based empirical mode decomposition badly, so that the decompose intrinsic mode function has no actual physical meaning. A new method for restraining the end effects of DEMD based on support vector machine （SVM） and window function is proposed, by using support vector machine for data extension on both ends of the signal, and adding a special window function after extension, and the continuation error is reduced. Simulation results and rolling bearing fault diagnosis examples show that the improved method can inhibit end effect effectively and has a higher accuracy of decomposition in rotating machinery fault diagnosis.