1. 东北大学 机械工程与自动化学院, 辽宁 沈阳 110819;
2. 沈阳机床集团有限责任公司 高档数控机床国家重点实验室, 辽宁 沈阳 110142
收稿日期:2017-04-21
基金项目:国家自然科学基金资助项目(51475087)。
作者简介:刘长福(1986-), 男, 内蒙古赤峰人, 东北大学博士研究生;
朱立达(1979-), 男, 吉林长春人, 东北大学教授, 博士生导师。
摘要:针对铣削过程中颤振频带不明显的问题, 采用变分模态分解(VMD)和快速傅里叶变换(FFT)相结合的方法来提取颤振频带, 为进一步提取颤振特征值奠定基础.为获得包含颤振频率的频带, 采用变切深侧铣薄壁件实验获取铣削力信号.提出结合FFT频谱来选择VMD中模态个数的方法, 并采用此方法对仿真信号和实验信号进行颤振频带提取, 结果表明VMD和FFT相结合的方法能有效提取铣削颤振频带.
关键词:侧铣颤振特征提取变分模态分解变切深薄壁件
Chatter Feature Extraction Method in Variable Cutting Depth Flank Milling Based on VMD and FFT
LIU Chang-fu1, ZHU Li-da1, QIU Jian2, LI Ming1
1. School of Mechanical Engineering & Automation, Northeastern University, Shenyang 110819, China;
2. State Key Laboratory of High Grade NC Machine Tools, Shenyang Machine Tool (Group) Co., Ltd., Shenyang 110142, China
Corresponding author: ZHU Li-da, E-mail: neulidazhu@163.com
Abstract: A chatter frequency band extracting method combining the variational mode decomposition (VMD) and fast Fourier transform (FFT) was proposed for extracting the obscure chatter frequency band in milling process, making good basis of further extracting the chatter feature value. To obtain the frequency band including chatter frequency, the variable cutting depth flank milling experiment was carried out and the forces were measured in the experiment. A method of choosing the number of modes in the VMD by combining the FFT spectrum was proposed. The simulation signal and experiment signal were extracted by the above method. The results show that the method of combining the VMD and FFT can effectively extract the chatter frequency bands in milling, which offers a new method to extract the chatter feature.
Key words: chatter in flank millingfeature extractionVMDvariable cutting depththin-walled workpiece
薄壁件刚度低、尺寸大, 在铣削过程中经常发生颤振.发生颤振时, 零件表面质量会受到严重影响.颤振特征提取是在线颤振检测的关键技术, 而实现颤振特征提取的前提是准确确定颤振频带.在颤振产生初期, 存在颤振特征不明显的问题, 研究高效、高精的早期颤振特征的提取方法对理论和实践有指导意义.
目前, 颤振特征提取方法有时域法、频域法、时频法.时频法由于能定位时间和频率, 在特征提取中被广泛应用[1-2].短时傅里叶变换(STFT)、小波变换(WT)、小波包(WPT)、同步压缩小波(SSWT)、经验模态小波[3]等方法在机械故障诊断和识别领域得到广泛应用[4-7].1998年, 黄锷等提出了经验模态分解方法(EMD), 并提出了集合经验模态分解方法(EEMD).EMD和EEMD方法在信号处理中得到了广泛的应用[8-10].2005年, Smith提出了局部均值分解(LMD).但是, 基于EMD的方法缺乏理论基础, 限制了其应用.
变分模态分解(VMD)是一种新的非递归式信号处理方法[11],Wang等[12]指出, VMD方法能更准确地提取特征.唐贵基等[13]采用基于包络谱特征因子(feature factor of envelope spectrum, EFF)的影响参数自动搜寻策略来选择模态个数和惩罚因子.钱林等[14]提出了利用互信息法来选择模态个数的方法, 然后利用形态学对信号进行降噪处理, 提取出滚动轴承的特征频率.VMD方法在特征提取中得到了广泛应用[15-18], 但是将VMD方法用于提取铣削过程中颤振特征则鲜少有文献报道.
为了验证VMD方法在铣削颤振频带识别中的有效性, 本文采用变切深侧铣薄壁件实验来获取铣削力, 对仿真信号和铣削实验信号分别用VMD和FFT相结合的方法来进行颤振频带提取.VMD的模态个数对识别结果有重要影响, 依据FFT频谱特点来选择模态个数, 结果表明VMD方法能准确确定铣削力信号的颤振频带, 为颤振特征提取提供了一种新的手段.
1 VMD方法的数学模型VMD方法的建模过程是:首先定义本征模态函数(IMF), 然后对变分问题进行构造, 最后求解变分问题.引用文献[17], 定义IMF为一个调幅-调频信号, 其表达式为
(1) |
(2) |
通过平方范数来估计各IMF的带宽, 如果将原始信号y(t)分解为K个IMF分量, 则对应的约束变分模型表达式为
(3) |
VMD中采用乘法算子交替方向法(ADMM)解决以上变分问题:
(4) |
(5) |
(6) |
2 仿真信号分析利用式(7)模拟含有颤振分量的铣削振动信号:
(7) |
图 1(Fig. 1)
图 1 原信号和VMD分解后各模态对比图Fig.1 Modes comparison between pre-VMD and the post-VMD signals |
图 2(Fig. 2)
图 2 频谱图Fig.2 Frequency spectrum (a)—原信号x(t)的频谱图;(b)—VMD后重组信号u的频谱图. |
原信号中x1(t)的频率w1=80 Hz, x2(t)的频率w2=20 Hz,以及x3(t)代表的颤振频带,均清晰地出现在重组信号中, 说明没有频率信息的遗漏.但重组信号u的频谱图中, 由噪声产生的高频信号相比原信号平坦许多, 说明VMD可以有效保留有用信息并去除噪声.
3 实验信号分析3.1 实验平台的搭建采用五轴数控机床DMU50实现变切深铣削, 铣刀直径D=10 mm, 齿数N=2, 薄壁件尺寸为100 mm×100 mm×5 mm, 见图 3.加工参数如下:径向切深ar=1 mm, 轴向切深从0 mm开始, 进给速度36 mm/min, 转速为2 000 r/min.采用测力仪Kistler 9257B采集x, y, z三向力信号, 整个切削过程在干铣削条件下进行.测力仪用压板固定在工作台上, 工件通过虎钳夹持.设置测力仪采样频率fs=7 000 Hz.
图 3(Fig. 3)
图 3 实验平台布局Fig.3 Milling experimental setup |
3.2 实验结果和讨论测量的铣削力及其频谱图如图 4所示.机床主频率SF=n/60, 机床刀具通过频率TPE=nNL/60, 其中L代表采集的通道数(比如采集的是xyz三个方向的切削力, L就等于3).
图 4(Fig. 4)
图 4 实验信号及其频谱图Fig.4 Experimental signal and frequency spectrum (a)—切削力;(b)—频谱图. |
依据颤振理论, 颤振频率接近于固有频率, 故颤振频率是主频的非整数倍.由图 4可以看出, 在频谱图中能找到4个频带.其中, 频带1, 2, 3属于主频率的谐振频带, 说明刀齿切入频率在力频谱中起主导作用[19], 频带1, 2, 3不包含颤振频率.而频带4不等于主频率的倍数, 说明系统结构模态频率在力频谱中起主导作用, 频带4中包含颤振频率.
将测量力信号经VMD处理, 因为图 4中整个频段分为4个频带, 所以取模态个数K=4, 惩罚因子α=2 000,对每一个IMF做FFT分析, 如图 5所示.图中可以看出, u1, u2, u3对应频带1, 2, 3, 而u4对应着颤振频带4.为验证VMD分解效果, 对VMD处理后的信号做时频分析, 如图 6所示.
图 5(Fig. 5)
图 5 VMD处理结果Fig.5 Result of VMD processing (a)—本征模态函数;(b)—频谱图. |
图 6(Fig. 6)
图 6 原信号和VMD处理后的信号时频图Fig.6 Time-frequency diagram of the row signal and signal after VMD (a)—原信号;(b)—VMD处理后. |
从图 6中可以看出, 原信号的频谱能量较均匀地分布在整个频域内, 时频图中频带1, 2, 3, 4的界限模糊, 能量分布不集中, 不利于后续的颤振特征值提取.而经VMD处理的时频图中, 4个频带所含频谱能量集中, 各频带带宽变窄, 颤振频带4被有效地提取出来.
4 结论1) 采用VMD结合FFT方法能有效提取铣削颤振频带, 方法简单、实用;
2) VMD方法中模态个数需预先设定, 本文通过FFT频谱选择模态个数, 从而有效地分解各模态并获得颤振频带.通过观察时频图可知颤振频带能量集中, 为提取颤振特征值奠定了的基础.
参考文献
[1] | Cao H, Zhou K, Chen X. Chatter identification in end milling process based on EEMD and nonlinear dimensionless indicators[J].International Journal of Machine Tools & Manufacture, 2015, 92: 52–59. |
[2] | Zhang C L, Li B, Chen B Q, et al. Weak fault signature extraction of rotating machinery using flexible analytic wavelet transform[J].Mechanical Systems & Signal Processing, 2015, 64/65: 162–187. |
[3] | Gilles J. Empirical wavelet transform[J].IEEE Transactions on Signal Processing, 2013, 61(16): 3999–4010.DOI:10.1109/TSP.2013.2265222 |
[4] | Cao H, Lei Y, He Z. Chatter identification in end milling process using wavelet packets and Hilbert-Huang transform[J].International Journal of Machine Tools and Manufacture, 2013, 69: 11–19.DOI:10.1016/j.ijmachtools.2013.02.007 |
[5] | Karam S, Teti R. Wavelet transform feature extraction for chip form recognition during carbon steel turning[J].Procedia CIRP, 2013, 12: 97–102.DOI:10.1016/j.procir.2013.09.018 |
[6] | Thakur G, Brevdo E, Fukar N S, et al. The synchros queezing algorithm for time-varying spectral analysis:robustness properties and new paleoclimate applications[J].Signal Processing, 2013, 93(5): 1079–1094.DOI:10.1016/j.sigpro.2012.11.029 |
[7] | Perez-Ramirez C A, Amezquita-Sanchez J P, Adeli H, et al. New methodology for modal parameters identification of smart civil structures using ambient vibrations and synchrosqueezed wavelet transform[J].Engineering Applications of Artificial Intelligence, 2016, 48: 1–12.DOI:10.1016/j.engappai.2015.10.005 |
[8] | Liu C F, Zhu L D, Ni C B. The chatter identification in end milling based on combining EMD and WPD[J].The International Journal of Advanced Manufacturing Technology, 2017, 91(9/10/11/12): 3339–3348. |
[9] | Ji Y, Wang X, Liu Z, et al. EEMD-based online milling chatter detection by fractal dimension and power spectral entropy[J].The International Journal of Advanced Manufacturing Technology, 2017, 92(1/2/3/4): 1185–1200. |
[10] | Wang W C, Chau K W, Xu D M, et al. Improving forecasting accuracy of annual runoff time series using ARIMA based on EEMD decomposition[J].Water Resources Management, 2015, 29(8): 2655–2675.DOI:10.1007/s11269-015-0962-6 |
[11] | Dragomiretskiy K, Zosso D. Variational mode decomposition[J].IEEE Transactions on Signal Processing, 2014, 62(3): 531–544.DOI:10.1109/TSP.2013.2288675 |
[12] | Wang Y, Markert R, Xiang J, et al. Research on variational mode decomposition and its application in detecting rub-impact fault of the rotor system[J].Mechanical Systems & Signal Processing, 2015, 60/61: 243–251. |
[13] | 唐贵基, 王晓龙. 变分模态分解方法及其在滚动轴承早期故障诊断中的应用[J].振动工程学报, 2016, 29(4): 638–648. ( Tang Gui-ji, Wang Xiao-long. Variational mode decomposition method and its application in the rolling bearing fault diagnosis[J].Journal of Vibration Engineering, 2016, 29(4): 638–648.) |
[14] | 钱林, 康敏, 傅秀清, 等. 基于VMD的自适应形态学在轴承故障诊断中的应用[J].振动与冲击, 2017, 36(3): 227–233. ( Qian Lin, Kang Min, Fu Xiu-qing, et al. The adaptive morphology in the application of the bearing fault diagnosis based on VMD[J].Journal of Vibration Engineering, 2017, 36(3): 227–233.) |
[15] | Wang Y, Markert R, Xiang J, et al. Research on variational mode decomposition and its application in detecting rub-impact fault of the rotor system[J].Mechanical Systems & Signal Processing, 2015, 60/61: 243–251. |
[16] | Abdoos A A, Mianaei P K, Ghadikolaei M R. Combined VMD-SVM based feature selection method for classification of power quality events[J].Applied Soft Computing, 2016, 38: 637–646.DOI:10.1016/j.asoc.2015.10.038 |
[17] | Aneesh C, Kumar S, Hisham P M, et al. Performance comparison of variational mode decomposition over empirical wavelet transform for the classification of power quality disturbances using support vector machine[J].Procedia Computer Science, 2015, 46: 372–380.DOI:10.1016/j.procs.2015.02.033 |
[18] | 吕中亮.基于变分模态分解与优化多核支持向量机的旋转机械早期故障诊断方法研究[D].重庆: 重庆大学, 2016. ( Lyu Zhong-liang.A novel method for mechanical fault diagnosis based on variational mode decomposition and multikernel support vector machine[D].Chongqing: Chongqing University, 2016.http://cdmd.cnki.com.cn/Article/CDMD-10611-1016765612.htm) |
[19] | 朱立达, 王宛山, 李鹤, 等. 正交车铣偏心加工三维颤振稳定性的研究[J].机械工程学报, 2011, 47(23): 186–192. ( Zhu Li-da, Wang Wan-shan, Li He, et al. Research on 3D chatter stability of eccentricity machining in orthogonal turn-milling[J].Journal of Mechanical Engineering, 2011, 47(23): 186–192.) |