Abstract: To improve the processing quality of products,the chatter stability lobes diagram was obtained with tests,and vibrate signals of CNC milling were collected to ensure the accuracy.The methods of wavelet packet transform and Hilbert-Huang transform were integrated to extract the characteristic value from two aspects of energy frequency distribution and statistical distribution,in which the influence of environmental noise interference could be reduced effectively by taking the wave let packet was as a pre-process,and the precision of EMD decomposition could be improved.The chatter identification model was established with Fuzzy Support Vector Machine (FSVM),and the vibration signal was differentiated into stable signal,weak chatter signal,chatter signal and tool-worn signal.The result of test indicated that the proposed model had fine ability of identifying and diagnosis in the field of CNC milling,and the prediction accuracy of the model could be achieved to 97.3%.A new way for detecting chatter of CNC milling accurately was also provided.

Key words: chatter, identification, wavelet packet, Hilbert-Huang transform, support vector machine, fuzzy, milling

摘要： 为了提高产品加工质量,根据试验测得铣削系统颤振稳定域,制定并采集数控铣削振动信号,以保证采集信号的准确性|融合小波包变换与希尔伯特黄变换,从能量频域分布与幅值概率统计分布两方面提取信号特征值,其中小波包降噪作为信号前置处理能有效降低环境噪声干扰的影响,提高经验模式分解的精度;建立基于模糊支持向量机的颤振诊断模型,将振动信号分为平稳铣削信号、微弱颤振铣削信号、颤振铣削信号及刀具磨损铣削信号。实验结果表明,该模型具有良好的铣削振动信号辨识与诊断能力,预测准确率达97.3%,为数控铣削加工振动信号的准确辨识与诊断提供了一种新方法。

关键词: 颤振, 诊断, 小波包, 希尔伯特黄变换, 支持向量机, 模糊, 铣削