| 159 | 3 | 1290 |
| 下载次数 | 被引频次 | 阅读次数 |
为了实现智能化轧制仿真过程，提出了将卷积神经网络(CNN)与有限元分析相结合的方法，利用机器代替人工选择出最佳的轧制仿真结果，并确定最优的孔型参数.首先，将CNN用于轧件道次分类，可发现“卷积神经网络+图像预处理”测试模型精度最高，为91.84%～96%;其次，使用5种评估方法对比轧制仿真结果与理想仿真结果后可得出，均方根误差和感知哈希对仿真结果更为敏感；最后，将孔型参数朝轧制仿真结果变优方向进行修正，对比仿真结果与修正前结果后发现，第2次修正更为理想.本研究中实现了CNN对棒材轧制道次分类，利用机器识别出更优的轧制仿真结果，并确定了更适合的孔型参数，在降低人力、物力消耗，以及实现仿真智能化等方面具有重要的理论意义和实际应用价值.
Abstract：In order to realize the intelligent rolling simulation process, the convolutional neural networks(CNN)and finite element analysis are combined to select the best rolling simulation results and determine the optimal pass parameters by machine instead of manual work. First of all, CNN is used for the classification of rolling pass, it is found that the test model “Convolutional neural network + image preprocessing” has the highest accuracy, which ranges from 91.84% to 96%. Secondly, five evaluation methods are used to compare the rolling simulation results with the ideal simulation results, and the RMS error and perceptual hash are more sensitive to the simulation results. Finally, the pass parameters are adjusted in the direction of the optimization of the rolling simulation results, and the simulation results are compared with the pre-adjustment results, and the second adjustment is more ideal. In this study, CNN classification of bar rolling passes has been realized, better rolling simulation results have been identified by the machine, and more suitable pass parameters have been determined. This method has important theoretical significance and practical application value in reducing manpower and material consumption and realizing intelligent simulation.
[1] 林朋雨，郭杰.基于FPGA的卷积神经网络加速优化方法[J].计算机仿真，2022,39(7):371-374.(Lin Pengyu,Guo Jie.FPGA-based accelerated optimization method of convolutional neural network[J].Computer Simulation,2022,39(7):371-374.)
[2] 孔玮，刘云，李辉，等.基于图卷积网络的行为识别方法综述[J].控制与决策，2021,36(7):1537-1546.(Kong Wei,Liu Yun,Li Hui,et al.A survey of action recognition methods based on graph convolutional network[J].Control and Decision,2021,36(7):1537-1546.)
[3] 王上，唐欢容.一种基于混合粒子群优化算法的深度卷积神经网络架构搜索方法[J].计算机应用研究，2023,40(7):2019-2024.(Wang Shang,Tang Huanrong.Deep convolutional neural architecture search method based on hybrid particle swarm optimization algorithm[J].Application Research of Computers,2023,40(7):2019-2024.)
[4] 杨健，吴思炜.基于机器学习的钢铁轧制过程性能预测[J].钢铁，2021,56(9):1-9.(Yang Jian,Wu Siwei.Property prediction of steel rolling process based on machine learning[J].Iron & Steel,2021,56(9):1-9.)
[5] 李维刚，刘玮汲，谢璐，等.基于图卷积网络的热轧带钢轧制力预测[J].钢铁，2023,58(3):89-96.(Li Weigang,Liu Weiji,Xie Lu,et al.Rolling force prediction of hot rolled strip by graph convolutional networks[J].Iron & Steel,2023,58(3):89-96.)
[6] 刘雨婷，郑洲顺，张帆，等.基于DEFORM-3D的钼板多道次轧制仿真研究[J].中国体视学与图像分析，2020,25(2):157-165.(Liu Yuting,Zheng Zhoushun,Zhang Fan,et al.Simulation study onmultipass rolling of molybdenum plate based on DEFORM-3D[J].Chinese Journal of Stereology and Image Analysis,2020,25(2):157-165.)
[7] 王振华，刘元铭，王涛，等.铜/铝/铜复合板异步轧制变形行为有限元分析[J].锻压技术，2022,47(7):175-183.(Wang Zhenhua,Liu Yuanming,Wang Tao,et al.Finite element analysis on deformation behavior of Cu/Al/Cu clad sheet during asynchronous rolling[J].Forging & Stamping Technology,2022,47(7):175-183.)
[8] He J X,Wang X D,Song Y F,et al.A multiscale intrusion detection system based on pyramid depthwise separable convolution neural network[J].Neurocomputing,2023,530:48-59.
[9] Deng J F,Sun J,Peng W,et al.Imbalanced multiclass classification with active learning in strip rolling process[J].Knowledge-Based Systems,2022,255:109754.
[10] Ishida M,Kaneko N,Sumi K.MOJI:Character-level convolutional neural networks for malicious obfuscated JavaScript inspection[J].Applied Soft Computing,2023,137:110138.
[11] 丁敬，方实年，余延庆，等.国际先进棒材轧钢技术综述[J].天津冶金，2021(2):54-57.(Ding Jing,Fang Shinian,Yu Yanqing,et al.The review of international advanced technology on bar rolling[J].Tianjin Metallurgy,2021(2):54-57.)
[12] 王楷，梅瑞斌.“人工智能+有限元” 模型在轧制领域的研究进展[J].材料导报，2022,36(13):137-148.(Wang Kai,Mei Ruibin .On the research and application of “artificial intelligence plus finite element” models in the field of rolling[J].Materials Reports,2022,36(13):137-148.)
[13] 谷向磊，黄长清，蔡央，等.智能算法优化轧制规程的优缺点及发展趋势[J].热加工工艺，2018,47(21):1-6.(Gu Xianglei,Huang Changqing,Cai Yang,et al.Advantages and disadvantages of intelligent algorithm for optimizing rolling schedule and its development trend[J].Hot Working Technology,2018,47(21):1-6.)
[14] 谢璐阳，夏兆君，朱少华，等.基于卷积神经网络的图像识别过拟合问题分析与研究[J].软件工程，2019,22(10):27-29.(Xie Luyang,Xia Zhaojun,Zhu Shaohua,et al.Analysis and research of overfitting of image recognition based on convolutional neural networks[J].Software Engineering,2019,22(10):27-29.)
[15] 刘巨升，于红，杨惠宁，等.基于多核卷积神经网络(BERT+Multi-CNN+CRF)的水产医学嵌套命名实体识别[J].大连海洋大学学报，2022,37(3):524-530.(Liu Jusheng,Yu Hong,Yang Huining,et al.Recognition of nested named entities in aquature medicine based on multi-kernel convolution(BERT+Multi-CNN+CRF)[J].Journal of Dalian Ocean University,2022,37(3):524-530.)
[16] 吴蓓，肖黎.基于卷积神经网络的农作物智能图像识别分类研究[J].农机化研究，2023,45(12):20-23.(Wu Bei,Xiao Li.Research on crop intelligent image recognition and classification based on convolution neural network[J].Journal of Agricultural Mechanization Research,2023,45(12):20-23.)
[17] 李文华，牛国波，刘羽佳，等.基于多通道卷积神经网络的液压系统多故障诊断[J].组合机床与自动化加工技术，2023(3):91-94.(Li Wenhua,Niu Guobo,Liu Yujia,et al.Multi-fault diagnosis of hydraulic system based on multi-channel convolutional neural network[J].Modular Machine Tool & Automatic Manufacturing Technique,2023(3):91-94.)
[18] 刘磊，李舜酩，陆建涛.基于卷积神经网络的旋转传动部件故障诊断综述[J].机械设计，2022,39(10):1-8.(Liu Lei,Li Shunming,Lu Jiantao.Summary of fault diagnosis of rotating transmission components based on convolutional neural network[J].Journal of Machine Design,2022,39(10):1-8.)
[19] 刘栋，杨辉，姬少培，等.基于多模型加权组合的文本相似度计算模型[J].计算机工程，2023,49(10):97-104.(Liu Dong,Yang Hui,Ji Shaopei,et al.Text similarity computing model based on weighted combination of multiple models[J].Computer Engineering,2023,49(10):97-104.)
[20] 刘婉婷.基于张量分解的图像哈希算法[D].桂林：广西师范大学，2022.(Liu Wanting.Image hashing algorithms based on tensor decomposition[D].Guilin:Guangxi Normal University,2022.)
基本信息:
DOI：10.14186/j.cnki.1671-6620.2024.04.011
中图分类号:TG335.62;TP183;TB115
引用信息:
[1]张芳萍,张宏政,贾怀博,等.卷积神经网络与有限元分析在棒材轧制孔型参数仿真中的应用[J].材料与冶金学报,2024,23(04):379-385.DOI:10.14186/j.cnki.1671-6620.2024.04.011.
基金信息:
山西省先进钢铁材料重点科技创新平台项目(201805D115061-2); 太原科技大学研究生教育创新项目(XCX212108)