基于遗传算法和神经网络的3D增材印花工艺参数优化

doi:10.13475/j.fzxb.20180604807

摘要/Abstract

摘要：

针对3D增材印花工艺中刮刀压力、刮印速度、刮刀角度和油墨黏度等参数的组合对印花质量存在较大影响,但实际生产中各工艺参数组合无法实现最优这一问题,利用附加动量法改进下的BP神经网络构建3D增材印花工艺模型,通过实验参数对模型进行训练,确定工艺参数和印花质量间的非线性关系。利用遗传算法对该非线性函数进行极值寻优,从而得到3D增材印花工艺的最优参数组合:印花压力为4 800N,刮印角度为18°,刮印速度为400 mm/s,油墨黏度为170.5 Pa·s,该模型预测误差基本稳定在0.01范围之内。利用优化前后的参数进行对比实验,结果证明该算法可以实现3D增材印花工艺的质量预测和参数寻优,从而提升印花质量,缩短产品开发时间。

关键词: 3D增材印花工艺, BP神经网络, 遗传算法, 参数优化, 质量预测

Abstract:

Aiming at the problem that the combination of parameters such as blade pressure, squeegee speed, blade angle and ink viscosity in the 3D additive printing process has a great influence on the printing quality, but the combination of various process parameters in actual production cannot be optimal, the BP neural network improved by the additional momentum method was adopted to construct a 3D additive printing process model. The model was trained by the experimental parameters to determine the nonlinear relationship between process parameters and printing quality. The genetic algorithm was adopted to optimize the nonlinear function to achieve the optimal parameter combination of the 3D additive printing process: printing pressure of 4 800 N, squeezing angle of 18 degrees, squeezing speed of 400 mm/s and ink viscosity of 170.5 Pa·s, The model prediction error is basically stable within the range of 0.01. The comparison experiments were carried out using the parameters before and after optimization. The experimental results show that the algorithm can realize the quality prediction and parameter optimization of 3D additive printing process, thereby improving the printing quality and shortening the product development time.

Key words: 3D additive screen printing process, BP neural network, genetic algorithm, parameter optimization, quality prediction

中图分类号:

TS194.49

王晓晖, 刘月刚, 孟婥, 孙以泽. 基于遗传算法和神经网络的3D增材印花工艺参数优化[J]. 纺织学报, 2019, 40(11): 168-174.

WANG Xiaohui, LIU Yuegang, MENG Zhuo, SUN Yize. Optimization of process parameters for 3D additive screen printing based on genetic algorithm and neural network[J]. Journal of Textile Research, 2019, 40(11): 168-174.

图/表 8

图1

图2

图3

表1

3D增材印花机印花质量数据"

样本编号	印花压力/N	刮印角度/(°)	刮印速度/ (mm·s^-1)	油墨黏度/ (Pa·s)	油墨转移率	样本编号	印花压力/ N	刮印角度/(°)	刮印速度/ (mm·s^-1)	油墨黏度/ (Pa·s)	油墨转移率
1	4 500	20	350	172.1	0.386	55	6 500	10	350	180.7	0.219
2	5 000	20	350	172.1	0.267	56	6 500	10	500	180.7	0.363
3	5 500	20	350	172.1	0.386	57	5 000	20	350	180.7	0.298
4	5 500	10	350	111.4	0.327	58	5 500	20	350	180.7	0.275
5	6 500	10	350	111.4	0.279	59	5 500	20	350	180.7	0.275
6	5 000	10	350	111.4	0.357	60	5 500	10	350	180.7	0.342
7	4 500	20	350	121.4	0.368	61	6 500	20	500	180.7	0.401
8	5 500	20	350	121.4	0.345	62	6 500	20	500	180.7	0.405
9	5 500	20	350	121.4	0.367	63	6 500	20	500	180.7	0.332
10	6 500	10	350	121.4	0.289	64	6 500	10	700	180.7	0.34
11	6 500	10	350	121.4	0.289	65	6 500	10	700	180.7	0.372
12	6 500	10	500	121.4	0.319	66	6 500	25	350	180.7	0.275
13	4500	20	350	155.0	0.459	67	6 500	25	350	180.7	0.249
14	5 500	20	350	155.0	0.317	68	6 500	5	350	180.7	0.281
15	5 500	20	350	155.0	0.317	69	6 500	5	700	180.7	0.212
16	6 500	10	350	155.0	0.376	70	6 500	5	700	180.7	0.212
17	6 500	10	350	155.0	0.376	71	6 500	20	500	180.7	0.265
18	6 500	10	500	155.0	0.321	72	6 500	20	500	180.7	0.315
19	4500	20	350	122.1	0.286	73	6 500	10	500	180.7	0.324
20	5 500	20	350	122.1	0.318	74	6 000	10	500	180.7	0.317
21	5 500	20	350	122.1	0.318	75	5 500	5	500	180.7	0.296
22	5 500	10	350	142.1	0.316	76	6 500	5	500	180.7	0.288
23	5 500	10	350	142.1	0.316	77	6 500	20	400	180.7	0.366
24	5 500	10	500	180.7	0.229	78	6 500	20	400	180.7	0.277
25	6 500	20	350	180.7	0.256	79	6 500	10	500	180.7	0.274
26	6 500	20	350	180.7	0.217	80	6 500	10	700	180.7	0.294
27	6 500	20	350	180.7	0.232	81	5 500	5	700	180.7	0.318
28	6 500	10	350	180.7	0.211	82	5 500	5	700	180.7	0.318
29	6 500	10	350	180.7	0.207	83	5 500	20	500	180.7	0.343
30	6 500	20	350	180.7	0.215	84	5 500	20	500	180.7	0.372
31	6 500	20	350	180.7	0.278	85	5 500	20	500	180.7	0.361
32	6 500	20	350	180.7	0.273	86	5 000	10	500	180.7	0.392
33	6 000	10	350	180.7	0.232	87	5 000	10	500	180.7	0.381
34	6 000	10	350	180.7	0.216	88	6 500	20	500	180.7	0.329
35	6 000	20	350	180.7	0.198	89	6 500	20	500	180.7	0.335
36	7 000	20	350	180.7	0.187	90	6 500	20	500	180.7	0.321
37	6 000	20	350	180.7	0.237	91	6 000	5	500	180.7	0.265
38	6 000	10	350	180.7	0.256	92	6 000	5	700	180.7	0.241
39	5 500	10	350	180.7	0.263	93	6 500	20	500	180.7	0.337
40	5 500	10	350	180.7	0.263	94	6 500	20	500	180.7	0.348
41	5 500	10	350	180.7	0.237	95	6 500	20	500	180.7	0.329
42	7 500	20	350	180.7	0.276	96	6 000	5	500	180.7	0.348
43	7 500	20	350	180.7	0.289	97	6 500	5	500	180.7	0.373
44	7 500	20	350	180.7	0.231	98	6 500	20	500	180.7	0.424
45	7 500	10	350	180.7	0.314	99	6 500	20	500	180.7	0.424
46	7 500	10	350	180.7	0.316	100	6 500	20	500	180.7	0.424
47	6 500	20	350	180.7	0.329	101	6 500	5	500	180.7	0.411
48	6 000	20	350	180.7	0.312	102	6 500	5	500	180.7	0.411
49	6 500	20	350	180.7	0.314	103	4 000	20	400	180.7	0.277
50	6 500	10	350	180.7	0.305	104	4 000	20	400	180.7	0.226
51	6 500	10	500	180.7	0.297	105	4 000	20	400	180.7	0.338
52	6 500	20	350	180.7	0.312	106	4 500	5	400	180.7	0.347
53	6 500	20	350	180.7	0.312	107	4 500	5	400	180.7	0.347
54	6 500	20	350	180.7	0.299	108	4 000	20	400	180.7	0.259

表1

图4

图5

图6

图7

参考文献 13

[1]	徐建国. 中国印刷产业的现状及发展趋势[J]. 印刷工业, 2017(5):46-47.
	XU Jianguo. The current situation and development trend of China's printing industry[J]. Printing Industry, 2017(5):46-47.
[2]	NATALJA Fjodorova, MARJANA Novic. Searching for optimal setting conditions in technological processes using parametric estimation models and neural network mapping approach: a tutorial[J]. Analytica Chimica Acta, 2015,891:90-100. doi: 10.1016/j.aca.2015.06.020 pmid: 26388367
[3]	李婧伟, 唐正宁, 王志宏. 基于印制电路板精细丝网印刷过程的数学模型研究[J]. 包装工程, 2008,29(2):94-95.
	LI Jingwei, TANG Zhengning, WANG Zhihong. Study of the mathematical model for fine screen printing process of print circuit board[J]. Packaging Engineering, 2008,29(2):94-95.
[4]	付尧建, 钱军浩. 基于柔性版最大印刷压力的数字模型研究[J]. 包装工程, 2013(17):120-123.
	FU Yaojian, QIAN Junhao. Digital model of maximum printing pressure for flexographic plate[J]. Packaging Engineering, 2013(17):120-123.
[5]	徐洋, 朱治潮, 盛晓伟. 基于机器视觉的鞋面特征点自动识别改进方法[J]. 纺织学报, 2019,40(3):168-174.
	XU Yang, ZHU Zhichao, SHENG Xiaowei, et al. Improvement recognition method of vamp's feature points based on machine vision[J]. Journal of Textile Research, 2019,40(3):168-174.
[6]	李淑娟, 陈文彬. 基于神经网络和遗传算法的三维打印工艺参数优化[J]. 机械科学与技术, 2014,33(11):1688-1693.
	LI Shujuan, CHEN Wenbin. Optimization of the 3D printing parameters based on the neural networks and genetic algorithm[J]. Mechanical Science and Technology for Aerospace Engineering, 2014,33(11):1688-1693.
[7]	左硕. 运动鞋面机器印花工艺参数优化研究[D]. 上海:东华大学, 2017:27-30.
	ZUO Shuo. Optimization research of process parameters for machine screen printing on shoes-upper[D]. Shanghai: Donghua University, 2017:27-30.
[8]	许小颖, 周盼, 王宽. 基于神经网络和遗传算法的系泊优化设计[J]. 中国舰船研究, 2017,12(5):97-103.
	XU Xiaoying, ZHOU Pan, WANG Kuan. Mooring optimization design based on neural network and genetic algorithm[J]. Chinese Journal of Ship Research, 2017,12(5):97-103.
[9]	张德丰. MATLAB神经网络编程[M]. 北京: 化学工业出版社, 2011: 23-46.
	ZHANG Defeng. MATLAB Neural Network Programming[M]. Beijing: Chemical Industry Press, 2011: 23-46.
[10]	REBEKA Fijan, MICHELE Basile, ROMANO Lapasin, et al. Rheological properties of printing pastes and their influence on quality-determining parameters in screen printing of cotton with reactive dyes using recycled polysaccharide thickeners[J]. Carbohydrate Polymers, 2009,78:25-35.
[11]	JASPER J MICHELS, SUZANNE H P M de Winter, LAURENCE H G Symonds. Process optimization of gravure printed light-emitting polymer layers by a neural network approach[J]. Organic Electronics, 2009,10:1495-1504.
[12]	曾军亮, 王荣吉, 冯晓欣. 基于神经网络和遗传算法的工艺参数优化[J]. 塑料, 2013(5):106-109.
	ZENG Junliang, WANG Rongji, FENG Xiaoxin. Optimization of process parameters based on neural network and genetic algorithm[J]. Plastics, 2013(5):106-109.
[13]	WANG Yacheng, LI Peibo, SUN Zhijun, et al. A model of screen reaction force for the 3D additive screen printing[J]. Journal of The Textile Institute, 2018,109(8):1000-1007.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

[1]	谢子昂, 杜劲松, 赵国华. 衬衫吊挂流水线的自适应动态调度[J]. 纺织学报, 2020, 41(10): 144-149.
[2]	张晓侠, 刘凤坤, 买巍, 马崇启. 基于BP 神经网络及其改进算法的织机效率预测[J]. 纺织学报, 2020, 41(08): 121-127.
[3]	黄珍珍, 莫碧贤, 温李红. 基于遗传算法及仿真技术的服装生产流水线平衡[J]. 纺织学报, 2020, 41(07): 154-159.
[4]	郑小虎, 鲍劲松, 马清文, 周衡, 张良山. 基于模拟退火遗传算法的纺纱车间调度系统[J]. 纺织学报, 2020, 41(06): 36-41.
[5]	莫帅, 冯战勇, 唐文杰, 党合玉, 邹振兴. 基于神经网络和遗传算法的锭子弹性管性能优化[J]. 纺织学报, 2020, 41(04): 161-166.
[6]	黄淇, 周其洪, 张倩, 王绍宗, 范伟, 孙会丰. 基于系统布置设计-遗传算法的纱线浸染生产线布局优化[J]. 纺织学报, 2020, 41(03): 84-90.
[7]	张旭靖, 王立川, 陈雁. 基于遗传算法的服装缝制生产线平衡优化[J]. 纺织学报, 2020, 41(02): 125-129.
[8]	孟朔潘如如高卫东王静安周利军. 采用主目标进化遗传算法的织造排程研究[J]. 纺织学报, 2019, 40(08): 169-174.
[9]	周贤王英陈建能汪锐陶德华. 缫丝机卷绕机构参数优化及其试验验证[J]. 纺织学报, 2019, 40(06): 97-105.
[10]	查刘根, 谢春萍. 应用四层BP神经网络的棉纱成纱质量预测[J]. 纺织学报, 2019, 40(01): 52-56.
[11]	巫莹柱　单颖法　黄伯熹　林广茂　梁家豪　张晓利. 聚对苯二甲酸丙二醇酯与聚对苯二甲酸丁二醇酯混纺纤维的智能识别[J]. 纺织学报, 2018, 39(09): 169-175.
[12]	韩瑞叶杨瑞华薛元高卫东. 数码转杯纺的Stearns-Noechel配色模型[J]. 纺织学报, 2017, 38(12): 27-32.
[13]	张宁潘如如高卫东. 采用图像处理的织物缝纫平整度自动评估[J]. 纺织学报, 2017, 38(04): 145-150.
[14]	宋楚平李少芹. 应用改进遗传算法的自动配棉模型优化与应用[J]. 纺织学报, 2016, 37(09): 151-155.
[15]	谷有众高卫东卢雨正刘建立杨瑞华. 应用遗传算法优化支持向量回归机的喷气涡流纺纱线质量预测[J]. 纺织学报, 2016, 37(07): 142-148.