To achieve the automatic identification of the internal combustion engine valve clearance, a ne fault diagnosis method based on the wavelet packet time-frequency phase plan image and Local nonnegative matrix factor-ization was proposed. The method uses wavelet packet to generate the time-frequency phase plane image of ICE vi-bration spectrum and a new method of local non-negative matrix factorization is directly used for feature extraction, because of the traditional NMF non-orthogonal basis matrix result in data redundancy is large, and it is not condu-cive to subsequent recognition rate of issue. The application to the practical internal combustion fault diagnosis has revealed that the LNMF algorithm classification accuracy is better than traditional NMF algorithm.%为实现内燃机气门间隙的自动识别，提出了一种基于小波包时频相平面图及局部非负矩阵分解的内燃机故障诊断方法。该方法首先采用小波包生成内燃机振动谱时频相平面图，然后针对传统非负矩阵分解非正交的基矩阵导致数据冗余性较大、不利于后续识别率提高的问题，提出采用局部非负矩阵分解的方法对生成的图形进行局部非负矩阵分解，提取出图像的代数特征值并进行故障诊断。对内燃机4种状态下信号的小波包时频相平面图的特征提取和故障分类结果表明，基于小波包时频相平面图与局部非负矩阵分解的方法对内燃机气门间隙的故障诊断的有效性。
展开▼
