为了直接对内燃机振动谱图像进行诊断识别，提出一种基于改进变分模态分解（ＶＭＤ）、伪魏格纳时频分 析（ＰＷＶＤ）与局部非负矩阵分解（ＬＮＭＦ）的内燃机振动谱图像识别诊断方法-In order to direct the internal combustion engine vibration spectral image for diagnosis recognition is proposed based on the improv

对原来的EMD进行了改进，使得对信号的分解更为完美，极大的抑制了端点效应和模态混叠现象。(The improvement of the original EMD makes the signal decomposition more perfect and greatly suppresses the endpoint effect and modal aliasing.)

改进的变分模态分解算法，有效提取振动信号中故障频率。(The improved variational mode decomposition algorithm can effectively extract the fault frequency in vibration signals.)