局部波动特征分解(LOD)方法及其在机械故障诊断中的应用研究

提出了一种新的自适应时频分析方法- - 局部波动特征分解（Local oscillatory-characteristic decomposition，LOD）方法，该方法以信号本身的局部波动特征为基础，并采用微分、坐标域变换、分段线性变换等运算手段将信号分解为一系列瞬时频率具有物理意义的单一波动分量（Mono-oscillation component，MOC）。每一个MOC分量都代表了原始信号的一种波动特征，从而能反映原信号的本质特征，所以LOD方法特别适合处理具有非平稳和非线性特性的机械故障振动信号。然而，LOD方法还有理论问题需要研究与完善，因此，本项目拟对LOD方法的若干理论问题进行研究与完善，并对该方法在机械故障特征提取中的应用进行研究，研究解决用LOD方法提取滚动轴承、齿轮和转子系统各类振动信号故障特征问题，在此基础上建立一套基于LOD的机械故障诊断方法。

A new self-adaptive time-frequency analysis method named local oscillatory-characteristic decomposition (LOD) is proposed. This method is based on local oscillatory characteristics of signal itself, and it uses the operations including differential, coordinates domain transform and piecewise linear transform to decompose the signal into a series of mono-oscillation components (MOC) which instantaneous frequency has the physical meanings. Each of MOC represents a oscillatory characteristics of original signal, and thus reflects the intrinsic characteristics of original signal, so the LOD method is especially suitable for processing the mechanical fault vibration signals which have the characteristics of non-stationary and nonlinearity. Nevertheless, the LOD method has some theoretical problems need to study and perfect, therefore, this project intends to study and perfect some theoretical problems of the LOD method, and studies the application of this method in machinery fault characteristics extraction, thus solves the problem of extracting the fault characteristics from vibration signals of rolling bearing, gear and rotor system by LOD method, and based on this, a machinery fault diagnosis method based on LOD will be established.