| 基于核超限学习机的轴向柱塞泵故障诊断 |
| |
| 引用本文： | 曾祥辉,兰媛,黄家海,胡晋伟,魏晋宏,武兵.基于核超限学习机的轴向柱塞泵故障诊断[J].液压与气动,2018,0(1):61-64. |
| |
| 作者姓名： | 曾祥辉 兰媛 黄家海 胡晋伟 魏晋宏 武兵 |
| |
| 作者单位： | 太原理工大学机械工程学院, 山西太原 030024 |
| |
| 基金项目： | 国家自然科学基金(51405327);山西省科技成果转化与推广计划项目(20051002) |
| |
| 摘 要： | 由于柱塞泵内部结构复杂且结构之间相互耦合, 致使对其进行故障诊断的难度也随之增加。为了提高算法的可靠性和诊断速度, 将核函数与超限学习机结合的方法用于柱塞泵故障诊断。首先, 通过加速度计和流量计采集到泵在正常和不同故障工况下的振动和流量信号, 同时对其采用小波包分解进行去噪;然后提取了时域无量纲指标和小波包分解的频带能量值中最大频带能量和系统中流量计的流量值, 共8维特征向量;最后用核超限学习机对4种故障(滑靴磨损、配油盘磨损、中心弹簧失效、松靴)进行识别与诊断。结果表明, 将核超限学习机用于故障诊断, 相比于超限学习机和传统的智能诊断算法支持向量机、BP神经网络有明显的优势。
|
| 收稿时间： | 2017-10-10 |
Fault Diagnosis Based on Kernel Extreme Learning Machine for Axial Piston Pump |
| |
| ZENG Xiang-hui,LAN Yuan,HUANG Jia-hai,HU Jin-wei,WEI Jin-hong,WU Bing.Fault Diagnosis Based on Kernel Extreme Learning Machine for Axial Piston Pump[J].Chinese Hydraulics & Pneumatics,2018,0(1):61-64. |
| |
| Authors: | ZENG Xiang-hui LAN Yuan HUANG Jia-hai HU Jin-wei WEI Jin-hong WU Bing |
| |
| Affiliation: | School of Mechanical Engineering, Taiyuan University of Technology, Taiyuan, Shanxi 030024 |
| |
| Abstract: | The complex internal structure of the piston pump and the coupling between structures result in the increasing difficulty of the piston pumps fault diagnosis. In order to improve the reliability and diagnostic speed of the algorithm, the method of combining the kernel function and the extreme learning machine is used to diagnose the fault of piston pump. Firstly, the vibration and flow rate signal of the pump under normal and different fault conditions are collected by an accelerometer and a flowmeter, and the wavelet packet decomposition is used to remove the noise. Then, a total of 8-dimensional characteristic vectors are extracted, including the time domain dimensionless index, the maximum energy of band energy decomposed by the wavelet packet and the flow value of the flowmeter in the system. Finally, the kernel extreme learning machine is used to identify and diagnose 4 kinds of faults (slipper abrasion, valve plate abrasion, central spring failure, and loose slipper fault). The experimental results show that compared with the extreme learning machine, the traditional intelligent diagnosis algorithm support vector machine and the back propagation neural network, the kernel extreme learning machine has obvious advantages in fault diagnosis. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《液压与气动》浏览原始摘要信息 |
|
点击此处可从《液压与气动》下载全文 |
|
