基于SST时频图纹理特征的供输弹系统故障诊断  

Fault diagnosis of the ammunition supply system based on the texture featuresof SST time-frequency distribution image

作　　者：潘宏侠[1] 张玉学 PAN Hongxia;ZHANG Yuxue(Mechanical Engineering Institute,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学机械工程学院,太原030051

出　　处：《振动与冲击》2020年第6期132-137,175,共7页Journal of Vibration and Shock

基　　金：国家自然科学基金(51675491,51175480)。

摘　　要：对于供输弹系统早期故障中信号成分复杂,故障征兆难以识别的问题,提出了基于同步压缩变换(SST)时频图纹理特征的故障诊断方法。使用EEMD方法对供输弹系统振动信号进行预处理,对分解的分量进行相关系数运算,选取与原始信号相关系数大的前4层分量对信号进行重构,达到了一定的降噪效果;接着利用供输弹系统不同状态的信号通过同步压缩变换时频分析,得到反映不同运行状态的二维时频图像,并进行灰度化处理;利用灰度共生矩阵法与灰度梯度共生矩阵,对其进行纹理特征的提取,为与传统方法做对比,提取了信号经EEMD分解后,与原始信号相关系数大的前4层分量的能量百分比作为特征;使用基于核的模糊C均值聚类,对供输弹系统三种不同状态振动信号的图像纹理特征和能量百分比特征进行分类识别,并与模糊C均值聚类进行对比。实验结果表明,该方法能有效地对自动供输弹系统早期故障进行识别,且识别正确率达91.21%。For the complicated signal components in the early failure of the supply and delivery missile system,the failure symptoms are usually difficult to identify.Aiming at this,an intelligent fault diagnosis method was proposed based on the texture features of synchrosqueezing wavelet transform(SST)time-frequency distribution images.The EEMD method was adopted to preprocess the vibration signal of the projectile delivery system,and to calculate the correlation coefficient of the decomposed components.The first four layers with high correlation coefficient were selected to reconstruct the signal for reducing the noise effect.Then,the vibration signals of different states of the transmissive bomb system were dealt with by the time-frequency analysis with the synchrosqueezing wavelet transform to obtain two-dimensional time-frequency images reflecting different operating states,and the gray-scale processing was performed.The gray level co-occurrence matrix method and the gray gradient co-occurrence matrix were used to extract the texture features.In order to compare with the traditional method,the energy percentage of the first four layers with large correlation coefficient of the original signal was taken as a feature after the signal was decomposed by EEMD.The method of kernel-based fuzzy C-means clustering,was used to do the classification and recognition of the image texture features and the image features of three different state vibration signals of the supply and delivery bomb system respectively,and the results were compared with those of the fuzzy C-means clustering.The experimental results show that the method can effectively recognize the early failure of automatic missile systems and the recognition accuracy is 91.21%.

关 键 词：供输弹系统 同步压缩变换(SST) 纹理特征提取 模糊核聚类 故障诊断 

分 类 号：TH17[机械工程—机械制造及自动化]