机械轴承故障信号提取仿真研究  

Simulation Study on Fault Signal Extraction of Mechanical Bearing

作　　者：孙红星[1] 张洋[1] 

机构地区：[1]辽宁科技大学电子与信息工程学院,辽宁鞍山114051

出　　处：《计算机仿真》2016年第3期419-423,共5页Computer Simulation

基　　金：国家自然科学基金(71472081)

摘　　要：有效提取机械轴承故障信号中的特征频率是判断轴承故障类型的基础,由于轴承故障信号中夹杂了噪声成分,影响了对故障特征信息的提取。有效滤除噪声成分的同时保留有用信号成分是提取故障信息的关键,所以研究了一种小波阈值方法与补充的总体平均经验模态分解(CEEMD)相结合的故障提取方法。首先介绍了一种小波阈值函数,同时引人了针对振动信号峭度值的阈值策略,运用这种降噪方法对轴承故障信号进行降噪预处理,然后结合改进的EMD分解方法-CEEMD(补充的总体平均经验模态分解),对降噪后的滚动轴承的故障振动信号进行分解处理,利用相关系数法和峭度值法选取了最佳本征模式分量(IMF),最后对IMF分量进行包络谱变换提取了轴承的故障频率,仿真结果证明所探讨的方法提取的故障信息更加清晰准确。Extracting the characteristic frequency in the mechanical bearing fault signal effectively is the basis of the judgement about bearing fault types. The beating fault signal is mixed with noise components, which affects the extraction of the fault characteristic information. Filtering out noise components effectively while retaining the useful signal components is the key to extract the fault information, so a fault extraction method was studied based on wavelet threshold method combined with Complementary Ensemble Empirical Mode Decomposition （CEEMD）. First, a threshold wavelet threshold function was introduced. In the meantime, the threshold policy based on kurtosis value contraposing the vibration signal was introduced. This noise reduction method was used for de-noising on the bearing fault signal. Then, combined with the improved EMD decomposition approach, Complementary Ensemble Empirical Mode Decomposition （CEEMD） , the actual measured fault vibration signals of rolling bearing were processed for noise reduction and decomposition. The correlation coefficient method and kurtosis value method were used to select the best intrinsic mode functions （ IMF）, then extracting the bearing failure frequency in IMF component by the spectral envelope conversion. The simulation results prove that the fault information extracted by this method is more clear and accurate.

关 键 词：滚动轴承 振动信号 小波阈值 故障提取 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]