摘要
采用S变换对柴油机气阀机构8种状态下的缸盖表面振动信号进行分析处理,得到一系列振动信号的时频图像。选取一部分时频图像组建得到各种状态的标准时频图像后,再根据测试图像和标准图像之间的欧氏距离、绝对值距离和相似度3种指标对测试图像进行识别分类,从而将气阀机构的故障诊断转换为时频图像的分类研究。试验结果表明,采用这一方法可以取得较好的诊断结果,对时频图像进行5次平均后,根据相似度和欧氏距离进行分类的正确识别率可以达到99 4%以上。
Several typical faults of diesel engines valve train were simulated in this paper. The vibration acceleration signals, which were acquired from the cylinder head, were analyzed with S Transform and then a series of timefrequency images were obtained. Eight standard timefrequency images were built from these images. By using some indexes such as Euclid distance, absolute distance and similarity between testing images and standard images, the testing images were classified into eight kinds, which correspond to eight states of valve train. Then the fault diagnosis for valve train was changed to classify timefrequency images. The experimental results show that this method can recognize the state of valve train correctly. Based on the indexes of similarity and Euclid distance, the rate of correct recognition can be as high as 994% when the images are averaged 5 times before classifying.
出处
《内燃机学报》
EI
CAS
CSCD
北大核心
2003年第4期271-275,共5页
Transactions of Csice
基金
863"计划资助项目(2001AA411310)。