基于振动信号的齿轮泵故障诊断  

Gear Pump Fault Diagnosis based on Vibration Signals

作　　者：何庆飞 陈小虎 王旭平 喻春明 张宁[1] HE Qingfei;CHEN Xiaohu;WANG Xuping;YU Chunming;ZHANG Ning(School of Mechanical Engineering, Xijing University, Xi’an 710123, China;School of Operational Support, Rocket Force Engineering University, Xi’an 710025, China)

机构地区：[1]西京学院机械工程学院,西安710123 [2]火箭军工程大学作战保障学院,西安710025

出　　处：《噪声与振动控制》2019年第4期195-199,223,共6页Noise and Vibration Control

基　　金：国家科技重大专项基金资助项目(2017ZX04011010);国防预研基金资助项目(9140A27020309JB4701)

摘　　要：进行齿轮泵故障诊断时,首先采用基于马氏距离的传感器通道选择方法选择最佳振动信号,然后采用多项式最小二乘法去除信号趋势项,再基于五点三次平滑法对信号进行平滑预处理,最后分别提取基于峭度的时域特征、小波包能量特征和经验模态分解特征,运用最小二乘支持向量分类机进行状态识别。应用实例表明在运用经验模态分解提取各频带能量作为特征参数时状态识别方法具有更高的识别率,能更有效识别齿轮泵的状态。Sensor channel selecting method based on Mahalanobis distance is proposed for gear pump fault diagnosis. In order to reduce the noise pollution and improve signal-to-noise ratio, the polynomial trend eliminating and smooth preprocessing are utilized to pretreat the vibration signals of the hydraulic gear pump. Finally, time domain feature based on kurtosis, wavelet packet energy feature and empirical mode decomposition character are extracted respectively. The proposed fault diagnosis method is applied for the pattern recognition of the gear pump. The experimental results show that, the method of LS-SVC fault diagnosis based on empirical mode decomposition character has higher recognition accuracy.

关 键 词：振动与波 齿轮泵 经验模态分解 小波包分析 最小二乘支持向量分类机 状态识别 

分 类 号：TH165.3[机械工程—机械制造及自动化] TN911.7[电子电信—通信与信息系统]