基于离散广义S变换与双向二维主成分分析的内燃机故障诊断

摘要

For the problems of fault diagnosis of I.C.engines,a method was proposed,which is consisted of discrete generalized S-transformation and TD-2DPCA.First of all,vibration spectrum im-ages of cylinder head vibration signals were generated by discrete generalized S transform.Secondly, image matrix was bidirectional compressed by TD-2DPCA to reduce sizes of feature coefficient matrix effectively.Lastly,these feature matrixes obtained from image projects were served as enters of nea-rest neighbor classifier,which was used to achieve fault types'division.The method was applied to di-agnosis example of the vibration signals of valve mechanism eight operating modes,comparisons of different time-frequency characterizations and feature extraction methods were carried out.The results show that the proposed method provides a new way for fault diagnosis of I.C.engines.%针对内燃机气阀机构的故障诊断问题,提出一种将离散广义 S 变换和双向二维主成分分析(TD-2DPCA)相结合的诊断方法.该方法首先利用离散广义S变换将内燃机缸盖振动信号生成振动谱图像,然后利用TD-2DPCA对图像进行特征提取,有效减小特征系数矩阵的维数,最后,通过最近邻分类器进行分类识别.将该方法应用于内燃机气阀机构8种工况的诊断实例中,对比不同时频表征及特征提取方法的计算效率和识别精度,结果表明该方法可为内燃机故障诊断提供一条新途径.