基于混合插值与多项式修正的局部特征尺度分解端点效应抑制方法

Mothod for restraining the end-effect of local characteristic-scale decomposition based on the mixed interpolation and polynomial correction

局部特征尺度分解是一种有效的自适应分解方法,广泛地应用于轴承、齿轮等旋转机械的故障诊断中。为了提高分解质量,给出了一种混合插值与多项式修正的端点效应抑制方法。对端点邻域数据段,采用线性方法由均值点插值均值曲线;将端点邻域插值的均值曲线与中间数据段合并,按照三次样条方法插值形成均值曲线;在端点邻域,采用多项式拟合方法修正均值曲线,保证所得分量的光滑性。通过仿真数据和轴承内圈故障振动数据的分析,验证了方法的有效性。

The local characteristic-scale decomposition is an effective adaptive decomposition method, which has been widely applied in the diagnosis of rotating mechanical faults, such as bearing and gear faults. In order to improve the decomposition quality, an end-effect suppression method was given based on the mixed interpolation and polynomial correction. First, the data segment of the endpoint neighborhood was linearly interpolated by means of mean point data. Next, the mean curve of the endpoint neighborhood was combined with the middle data segment, and the mean curve was formed according to the third order spline interpolation. Then, the mean curve of the endpoint neighborhood was corrected by the cubic polynomial method, and the smoothness of the resulting component was guaranteed. The analysis results of the simulated data and real bearing fault data show that the method can restrain the end-effect of local characteristic-scale decomposition effectively.