基于对称极坐标方法和图像识别的柴油机曲轴轴承故障诊断  

Diagnosis of Diesel Engine Crankshaft Bearing Fault Based on Symmetric Polar Coordinates and Image Recognition

作　　者：张玲玲[1,2] 任金成[2] 封会娟[2] 朱忠奎[3] 肖云魁[2] 

机构地区：[1]军械工程学院火炮工程系,石家庄050003 [2]军事交通学院汽车工程系,天津300161 [3]苏州大学城市轨道交通学院,苏州215021

出　　处：《内燃机工程》2015年第4期144-149,共6页Chinese Internal Combustion Engine Engineering

基　　金：国家自然科学基金项目(50575063/E051301)

摘　　要：采用对称极坐标方法,将振动信号的时域波形不经过时频分析,直接转换为镜面对称雪花图像,通过提取图像区域几何特征参数质心和方向角反映不同磨损故障程度,最后采用模糊C均值聚类方法进行分类识别。在EQ6BT5.9型柴油机上进行了四种不同磨损工况下曲轴轴承故障诊断试验,诊断实例表明总体诊断正确率为86%。Symmetrical polar coordinates method was introduced to directly transform time domain waveform of vibration signal into mirror symmetry pattern of snowflake instead of through time-frequency analysis.The geometric feature parameters of image such as centroid position and orientation were extracted to reflect different abrasion degrees of crankshaft bearing,and the diagnosis of crankshaft bearing fault was realized by fuzzy C-means clustering recognition.The experimental results of the test of the EQ6BT5.9engine under four wearing conditions indicate that crankshaft bearing faults under all wear conditions can be diagnosed online with accuracy rate of 86%.

关 键 词：内燃机 柴油机 对称极坐标 镜面对称图像 故障诊断 

分 类 号：TK428[动力工程及工程热物理—动力机械及工程]