基于小波包分析的滚珠丝杠状态监测

Condition Monitoring of Ball Screw Based on Wavelet Packet

为了研究丝杠在不同寿命下的运行状态,准确判断出丝杠的性能退化程度,提出了一种基于小波包分析的丝杠运行状态监测方法。以丝杠性能退化试验平台为研究对象,安装三向加速度传感器,来采集丝杠的振动信号。通过对振动信号的进行小波包变换,采用能量谱筛选出包含丝杠状态信息的频率带,再从这些频率带中通过时域、频域分析的方法提取出敏感特征,最后运用机器学习的方法对特征进行分类,以达到监测丝杠运行状态的目的。试验结果表明通过该方法可以有效监测丝杠状态,平均识别准确率达到了95%以上。

To study the operating condition and performance degradation of screw under different life,a method was proposed to monitor the condition based on wavelet packet.By installing tri-axial acceleration sensors on test platform of screw performance degradation,a monitoring system was built so that we can acquire the vibration.The frequency bands which contain the information of screw condition can be screened out by energy spectrum after wavelet packet transformation of vibration signal,then selected most sensitive characteristic from these bands through time domain and frequency domain analysis.Using machine learning methods to classify the characteristic and finally identify the operation condition of ball screw.The experimental result showed that the method can efficiently monitor the condition of screw,the average identification rate is above 95%.