Study on Combined Prediction Model for Surface Roughness in Milling Process
-
摘要: 加工过程产生的粗糙度数据序列会包含多种特征，而单一的预测模型不能同时捕捉多种数据特征，难以提高预测精度。因此，从加工过程中粗糙度数据特征的复杂性出发，提出了一种基于支持向量机（SVM）和BP神经网络算法（BP）的组合预测模型，来同时捕捉数据的线性特征和非线性特征；在组合预测过程中为充分发挥两种预测算法的最佳性能，采用粒子群优化算法（PSO）对支持向量机的参数和BP神经网络中的权值进行优化。通过蠕墨铸铁的铣削实验，实现不同切削用量下的表面粗糙度精准预测，并与PSO-SVM、PSO-BP算法以及切削加工表面粗糙度理论模型进行对比，验证了该组合模型的优越性。Abstract: The roughness data sequence generated in machining process will contain a variety of features, and a single prediction model cannot simultaneously capture multiple data features, and it is difficult to improve the prediction accuracy. Therefore, a combined prediction model based on support vector machine (SVM) and BP neural network algorithm (BP) considering the complexity of data features in machining process is proposed, which can simultaneously capture the linear characteristics and nonlinearity of data features. In order to full play the two prediction algorithms, particle swarm optimization (PSO) is used to optimize the parameters of the support vector machine and the weights in the BP neural network. Through milling experiments, and comparing with PSO-SVM, PSO-BP algorithm and the model for surface roughness, the superiority of the combined model (PSO-SVM+PSO-BP) is verified.
-
Key words:
- combined model /
- prediction of surface roughness /
- parameters optimization /
- milling
-
表 1 实验切削用量及粗糙度值
实验组数 切削速度/ (m·min-1) 进给量/ (mm·min-1) 粗糙度实际值/μm 1 226 180 1.409 2 2 226 360 2.507 0 3 226 540 3.659 0 4 226 720 4.433 2 5 226 900 4.120 3 6 452 180 0.340 3 7 452 360 0.786 7 8 452 540 1.550 5 9 452 720 2.382 2 10 452 900 3.227 2 11 678 180 0.331 8 12 678 360 0.430 2 13 678 540 0.557 0 14 678 720 0.645 3 15 678 900 0.794 0 16 904 180 0.323 8 17 904 360 0.529 8 18 904 540 0.478 5 19 904 720 0.717 3 20 904 900 0.773 2 
下载: 导出CSV
表 2 PSO在SVM和BP算法中的参数设
参数名 SVM中的PSO参数值 BP中的PSO参数值 种群数量 20 20 迭代次数 200 100 学习因子(c1, c2) c1=1.5
c2=1.7c1=1.494 45
c2=1.494 45适应度函数 均方误差EMS 均方根误差ERMS
下载: 导出CSV
表 3 各模型对测试样本预测值及相对误差
样本号 表面粗糙度预测值/μm 相对误差ER/% PSO-SVM+PSO-BP PSO-SVM PSO-BP 理论模型 PSO-SVM+PSO-BP PSO-SVM PSO-BP 理论模型 13 0.544 6 0.597 4 0.624 9 0.758 0 2.223 7 7.246 0 12.187 9 36.094 4 14 0.697 5 0.597 4 0.703 5 0.945 2 8.094 5 7.429 1 9.013 1 47.866 1 15 0.794 9 0.694 5 0.780 0 1.140 6 0.119 5 12.528 3 1.768 0 43.656 5 16 0.323 9 0.423 9 0.333 0 0.214 1 0.030 4 30.909 5 2.834 4 33.868 3 17 0.497 4 0.597 4 0.439 9 0.372 8 6.119 5 12.752 0 16.970 3 29.634 5 18 0.491 8 0.578 6 0.603 1 0.515 6 2.784 2 20.916 4 31.682 3 7.755 6 19 0.716 3 0.617 0 0.690 1 0.649 0 0.141 4 13.989 7 3.787 0 9.519 3 20 0.773 3 0.672 9 0.698 4 0.775 8 0.010 1 12.978 2 9.670 9 0.341 4
下载: 导出CSV
表 4 各预测模型的评价指标计算结果
模型 EMAP/% ERSM/% PSO-SVM+PSO-BP 2.440 4 2.27 POS-SVM 14.843 6 8.56 PSO-BP 10.989 2 7.56 理论模型 26.092 0 19.32
下载: 导出CSV
-
[1] Upadhyay V, Jain P K, Mehta N K. In-process prediction of surface roughness in turning of Ti-6Al-4V alloy using cutting parameters and vibration signals[J]. Measurement, 2013, 46(1):154-160 doi: 10.1016/j.measurement.2012.06.002 [2] 孙林, 杨世元.基于最小二乘支持矢量机的成形磨削表面粗糙度预测及磨削用量优化设计[J].机械工程学报, 2009, 45(10):254-260 http://d.old.wanfangdata.com.cn/Periodical/jxgcxb200910039Sun L, Yang S Y. Prediction for surface roughness of profile grinding and optimization of grinding parameters based on least squares support vector machine[J]. Journal of Mechanical Engineering, 2009, 45(10):254-260(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/jxgcxb200910039 [3] Tsai Y H, Chen J C, Lou S J. An in-process surface recognition system based on neural networks in end milling cutting operations[J]. International Journal of Machine Tools and Manufacture, 1999, 39(4):583-605 doi: 10.1016/S0890-6955(98)00053-4 [4] 刘贵杰, 巩亚东, 王宛山.基于神经网络的磨削砂轮状态的在线监测[J].东北大学学报, 2002, 23(10):984-987 doi: 10.3321/j.issn:1005-3026.2002.10.018Liu G J, Gong Y D, Wang W S. On-line monitoring for grinding wheel states based on neural network[J]. Journal of Northeastern University, 2002, 23(10):984-987(in Chinese) doi: 10.3321/j.issn:1005-3026.2002.10.018 [5] 邓辉, 邓朝晖, 肖蓝湘.基于粒子群算法的钛合金高速外圆磨削参数优化[J].兵器材料科学与工程, 2017, 40(4):46-50 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=bqclkxygc201704011Deng H, Deng Z H, Xiao L X. Optimization of high-speed cylindrical grinding parameter for titanium alloy based on particle swarm optimization[J]. Ordnance Material Science and Engineering, 2017, 40(4):46-50(in Chinese) http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=bqclkxygc201704011 [6] Lu D W. Research on prediction method of profile grinding surface roughness based on PSO-LS SVM algorithm[J]. Advanced Materials Research, 2014, 926-930:3684-3687 doi: 10.4028/www.scientific.net/AMR.926-930.3684 [7] 阴艳超, 丁卫刚.切削加工表面粗糙度的多维多规则云预测方法[J].机械工程学报, 2016, 52(15):204-212 http://d.old.wanfangdata.com.cn/Periodical/jxgcxb201615027Yin Y C, Ding W G. A novel cloud model prediction for surface roughness based on multidimensional & multi-rules reasoning[J]. Journal of Mechanical Engineering, 2016, 52(15):204-212(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/jxgcxb201615027 [8] Kumar S, Batish A, Singh R, et al. A hybrid Taguchi-artificial neural network approach to predict surface roughness during electric discharge machining of titanium alloys[J]. Journal of Mechanical Science and Technology, 2014, 28(7):2831-2844 doi: 10.1007/s12206-014-0637-x [9] Krishna Mohana Rao G, Rangajanardhaa G, Hanumantha R D, et al. Development of hybrid model and optimization of surface roughness in electric discharge machining using artificial neural networks and genetic algorithm[J]. Journal of Materials Processing Technology, 2009, 209(3):1512-1520 doi: 10.1016/j.jmatprotec.2008.04.003 [10] Deris A M, Zain A M, Sallehuddin R. Hybrid GR-SVM for prediction of surface roughness in abrasive water jet machining[J]. Meccanica, 2013, 48(8):1937-1945 doi: 10.1007/s11012-013-9710-2 [11] Sahoo P, Pratap A, Bandyopadhyay A. Modeling and optimization of surface roughness and tool vibration in CNC turning of Aluminum alloy using hybrid RSM-WPCA methodology[J]. International Journal of Industrial Engineering Computations, 2017, 8(3):385-398 http://www.growingscience.com/ijiec/Vol8/IJIEC_2016_33.pdf [12] Hanafi I, Khamlichi A, Cabrera F M, et al. Optimization of cutting parameters in CNC turning operation using Taguchi design of experiments[J]. Science, 2011, 2(2):79-87 https://www.researchgate.net/publication/273218742_Optimization_of_cutting_parameters_in_CNC_turning_operation_using_Taguchi_design_of_experiments [13] 马峰, 张华, 曹华军.面向低能耗少切削液的多目标加工参数优化[J].机械工程学报, 2017, 53(11):157-163 http://d.old.wanfangdata.com.cn/Periodical/jxgcxb201711019Ma F, Zhang H, Cao H J. Multi-objective machining parameters optimization for low energy and minimum cutting fluid consumption[J]. Journal of Mechanical Engineering, 2017, 53(11):157-163(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/jxgcxb201711019 [14] 杨俊, 何辉波, 李华英, 等.基于切削力和表面粗糙度的干切削参数优化[J].西南大学学报, 2014, 36(12):187-192 http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=xnnydxxb201412030Yang J, He H B, Li H Y, et al. Optimization of dry cutting parameters based on cutting force and surface roughness[J]. Journal of Southwest University, 2014, 36(12):187-192(in Chinese) http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=xnnydxxb201412030 [15] 周长林, 钱志升, 王勤民, 等.基于PSO-SVM方法的电源线传导泄漏信号识别与还原[J].电子与信息学报, 2018, 40(9):2206-2211 http://d.old.wanfangdata.com.cn/Periodical/dzkxxk201809023Zhou C L, Qian Z S, Wang Q M, et al. Recognition and reconstruction of conduction leakage signal via power line based on PSO-SVM method[J]. Journal of Electronics & Information Technology, 2018, 40(9):2206-2211(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/dzkxxk201809023 [16] Rumelhart D E, McClelland J L, the PDP Research Group. Parallel distributed processing:explorations in the microstructure of cognition[M]. Cambridge MA:MIT Press, 1986 [17] 张捍东, 陶刘送.粒子群优化BP算法在液压系统故障诊断中应用[J].系统仿真学报, 2016, 28(5):1186-1190 http://d.old.wanfangdata.com.cn/Periodical/xtfzxb201605025Zhang H D, Tao L S. Application of PSO-BP algorithm in hydraulic system fault diagnosis[J]. Journal of System Simulation, 2016, 28(5):1186-1190(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/xtfzxb201605025 [18] 陈伟, 周曼, 叶家军, 等.建筑工程安全施工费费率测算的PSO-BP模型研究[J].中国安全科学学报, 2016, 26(5):146-151 http://d.old.wanfangdata.com.cn/Periodical/zgaqkxxb201605026Chen W, Zhou M, Ye J J, et al. PSO-BP modeling research on fee rate measurement of construction project safe construction cost[J]. China Safety Science Journal, 2016, 26(5):146-151(in Chinese) http://d.old.wanfangdata.com.cn/Periodical/zgaqkxxb201605026 [19] 张建伟, 江琦, 刘轩然, 等.基于PSO-SVM算法的梯级泵站管道振动响应预测[J].农业工程学报, 2017, 33(11):75-81 doi: 10.11975/j.issn.1002-6819.2017.11.010Zhang J W, Jiang Q, Liu X R, et al. Prediction of vibration response for pipeline of cascade pumping station based on PSO-SVM algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(11):75-81(in Chinese) doi: 10.11975/j.issn.1002-6819.2017.11.010 -
点击查看大图
图(3) / 表(4)
计量
- 文章访问数: 414
- HTML全文浏览量: 301
- PDF下载量: 24
- 被引次数: 0
