基于低秩稀疏分解算法的铣床齿轮箱故障诊断  

Fault Diagnosis of Milling Machine Gearbox Based on Low Rank Sparse Decomposition Algorithm

作　　者：于春霞[1] 张建国[2] 李明 YU Chun-xia;ZHANG Jian-guo;LI Ming(Department of Computer Science,Yellow River Institute of Science and Technology,He’nan Zhengzhou 450000,Chi-na;School of Mechanical Engineering,He’nan Polytechnic University,He’nan Zhengzhou 450000,China;He’nan Litian Tool Co.,Ltd.,He’nan Zhengzhou 450000,China)

机构地区：[1]黄河科技学院,计算机系,河南郑州450000 [2]河南理工大学,机械工程学院,河南郑州450000 [3]河南力天刀具有限公司,河南郑州450000

出　　处：《机械设计与制造》2023年第9期188-192,共5页Machinery Design & Manufacture

基　　金：河南省民办高等学校品牌专业建设项目(ZLG201903)。

摘　　要：铣床齿轮箱的安全运行对保证机械设备的效率具有重要的作用,其故障诊断复杂难控。传统形式算法只是从原始振动信号中进行字典原子学习,并未从本质层面分析特征信息物理结构特性。采用低秩稀疏分解算法,并进行BCD求解对齿轮箱故障诊断开展分析。研究结果表明:特征信号已淹没到了噪声中,能够对等间隔冲击特征进行准确识别,并使特征信号信噪比由-9.152增大为4.716。表明采用稀疏低秩算法能够滤除噪声干扰,从而高效识别瞬态冲击成分。经过3次迭代后特征信号发生了奇异值快速衰减现象,具有明显稀疏特性。低秩稀疏分解信号形成的包络谱,已经实现了所有干扰频率成分以及噪声成分的滤除效果,采用低秩稀疏分解算法能够实现齿轮箱局部故障的准确诊断。Safe operation of gear box plays an important role in ensuring the efficiency of mechanical equipment,and its fault di⁃agnosis is complex and difficult to control.The traditional algorithm only learns dictionary atoms from the original vibration sig⁃nals,but does not analyze the physical structure characteristics of feature information from the essential level.The low-rank sparse decomposition model and BCD solution were used to analyze the gearbox fault diagnosis.The results show that the charac⁃teristic signal has been submerged in the noise,and the corresponding interval impact feature can be accurately identified,and the SNR of the characteristic signal increases from-9.152 to 4.716.It is shown that the sparse low-rank algorithm can filter out the noise interference and identify the transient impact components efficiently.After three iterations,the singular values of the characteristic signal decay rapidly,and the signal has obvious sparse characteristics.The envelope spectrum formed by low-rank sparse decomposition signals has realized the filtering effect of all interference frequency components and noise components,and the low-rank sparse decomposition algorithm can realize the accurate diagnosis of local faults of the gearbox.

关 键 词：齿轮箱 故障诊断 稀疏分解 噪声 

分 类 号：TH16[机械工程—机械制造及自动化] TH137