基于最大提升格形态小波变换的齿轮故障特征提取  

Max-lifting morphological wavelet transform based gear fault feature extraction

作　　者：张培林[1] 李兵[1,2] 张英堂[1] 米双山[2] 刘东升[2] 

机构地区：[1]军械工程学院自行火炮教研室,石家庄050003 [2]军械工程学院导弹机电工程教研室,石家庄050003

出　　处：《仪器仪表学报》2010年第12期2736-2741,共6页Chinese Journal of Scientific Instrument

摘　　要：针对齿轮故障特征提取问题,提出了一种基于最大提升格形态小波变换的信号分解方法。最大提升格形态小波是在数学形态学和提升方案的基础上提出的一种非线性小波变换方法,具有信号局部极值保持和计算快速的优点。提出将最大提升格形态小波用于齿轮发生故障时所产生的非平稳、非线性振动加速度信号的分析,提取故障的特征信息。通过对仿真信号和实际的齿轮断齿故障信号的分析结果,证明了所采用方法的有效性。同时,与采用传统的线性小波分解分析结果相比,最大提升格形态小波变换能够在较高分解层次下十分有效地保留信号的冲击特征,能够利用较少的系数实现对故障信号的特征提取,而且最大提升格形态小波变换算法只涉及加减和取极大、极小运算,运算简单,执行高效,非常适于齿轮故障的在线监测和诊断。A new feature extraction technique based on max-lifting morphological wavelet（MLMW） decomposition is presented for gear fault diagnosis in this investigation.Max-lifting morphological wavelet is a novel nonlinear wavelet transform method based on mathematic morphology theory and lifting scheme.The most attractive characteristics of MLMW are low computation cost and the property to preserve important geometric information of the signal over a range of scales.MLMW algorithm is used to analyze the nonstationary and nonlinear acceleration signal and extract fault feature information of gear faults.Both simulated impulse signal and real gear fault vibration signal are employed to evaluate the presented method.Traditional linear wavelet transform method is also utilized to analyze the same signals;and the results for both methods are compared.Application results demonstrate that MLMW algorithm can preserve the impulse information at higher resolution level.Besides,MLMW algorithm involves only simple operations,such as addition,subtraction,maximum and minimum,and features simple calculation and high execution efficiency.Thus MLMW algorithm is highly suitable for the on line inspection and fault diagnosis of gear faults.

关 键 词：形态小波 最大提升格 齿轮 故障诊断 特征提取 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]