基于改进LMD的泵站摆度信号分析

局部均值分解(local mean decomposition, LMD)在分析非线性、非平稳信号时具有特有的分析能力,特别适合泵站机组摆度信号的分析.但是在信号分解为固有旋转分量的过程中,由于局域均值和包络估计函数在数据端点存在误差,会产生端点效应,严重时导致信号分解结果失真.针对这一问题,从全面考虑曲线幅值和几何形状相似性出发,提出基于灰色B型关联度和欧氏距离的端点效应抑制方法.为评价该端点效应抑制方法的抑制效果,提出分解前后的信号能量变化的评价标准.通过与原始局域均值分解算法、镜像映射法和波形匹配法等传统方法相比,验证该方法的有效性和优越性.仿真信号和泵站主轴摆度实测信号的应用表明,该方法能够有效地抑制端点效应,提高LMD分解过程中重构信号精度,更好地提取泵站机组摆度信号故障特征.

Analysis of pumping station swing signal based on improved LMD

The local mean decomposition(LMD)method presents unique analysis ability for processing nonlinear and non-stationary signals. It is especially suitable for signal analysis of pumping units. However, in the process of decomposition a non-stationary signal into several proper rotation components, due to the error of the local mean and envelope estimation function at the data endpoint. When the endpoint effect appears, which results in the distortion of signal decomposition in serious cases. To solve this problem, considering the curve amplitude and geometry shape similarity, an endpoint effect suppression method based on grey B correlation and Euclidean distance was proposed. In order to evaluate the effect of the endpoint suppression method, an evaluation criterion of the change of signal energy before and after the decomposition was defined. Compared with the original LMD method and the image mapping method and waveform matching method, the effectiveness and superiority of this proposed method is verified. The results of simulation and practical swing signal analysis of pumping station shows that this method can effectively suppress the endpoint effects of LMD, and improve the accuracy of reconstructed signal precision in the process of LMD decomposition, and better extract the fault signal characteristics of swing signal of pumping station unit.