论文部分内容阅读
经典的经验模式分解(EMD)方法通过求解信号的上下2条三次样条包络曲线的均值曲线,实现对原始信号的分解。但是对于非平稳、非线性信号,包络平均无法代替真正的局部平均。另外,基于包络平均的分解方法还会引入极值过冲和欠冲问题。利用B样条的良好局部性质直接计算信号的局部均值插值曲线,克服了三次样条包络方法在EMD分解中的不足。通过对旋转机械故障振动信号的分解处理,表明基于B样条局部均值插值曲线的经验模式分解方法得到的固有模式函数更符合信号的真实物理意义,分解结果更好。
The classical empirical mode decomposition (EMD) method solves the decomposition of the original signal by solving the mean curve of two cubic spline envelope curves. However, for non-stationary, non-linear signals, the envelope average can not replace the true local average. In addition, the envelope-based decomposition method also introduces extreme overshoot and undershoot problems. The local average interpolation curve of signals is directly calculated by using the good local properties of B-splines, which overcomes the shortcomings of the cubic spline envelope method in EMD decomposition. By decomposing the vibration signal of the rotating machinery, it is shown that the intrinsic mode function obtained by the empirical mode decomposition method based on the B-spline local mean interpolation curve is more in line with the real physical meaning of the signal, and the decomposition result is better.