基于图像识别的齿轮故障诊断方法研究  

Research of Gear Fault Diagnosis Based on Image Recognition

作　　者：魏维格 董辛旻[1] 韩捷[1] 郝旺身[1] 王成明[1] 

机构地区：[1]郑州大学机械工程学院,郑州450001

出　　处：《煤矿机械》2016年第5期166-169,共4页Coal Mine Machinery

基　　金：超深矿井提升装备的高提升能力创新结构设计(2014CB049402)

摘　　要：以圆柱齿轮为对象,提出了基于双谱分析和图像识别的齿轮箱故障诊断方法。通过小波包双谱分析生成故障信号的双谱图,并通灰度共生矩阵的2阶特征统计量来表征双谱图特征。对于这些特征统计量,通过相关度的大小进行分类从而得到故障模式识别的输入特征向量。通过支持向量机(SVM)模式识别算法对这些特征集合进行分类,从而识别出相应的齿轮故障类型。实验证明在具有噪声的情况下,该方法取得了比较理想的识别率,验证了基于图像识别的齿轮箱故障诊断方法的可行性。With the cylindrical gear as objects,puts forward the method about gearbox fault diagnosis based on bispectrum analysis and image recognition. By the bispectrum of wavelet packet analysis to generate fault signal of bispectrum image, and through the second order characteristic statistics of the gray symbiotic matrix to represent the characteristics？of bispectrum image. To get the fault pattern recognition input feature vectors we can classify the statistic classification by the size of the relevant characteristics. Through the support vector machine（SVM） pattern recognition algorithm to classify the characteristic set we can identify the corresponding fault gear type. The experiment proved that the method has obtained a good recognition rate in cases with noise and verify the feasibility of the method about gearbox fault diagnosis based on bispectrum analysis and image recognition.

关 键 词：齿轮箱故障诊断 小波包双谱分析 灰度共生矩阵 支持向量机(SVM) 

分 类 号：TH113.1[机械工程—机械设计及理论]