锻压技术

基于神经网络航空发动机曲轴加工工艺设计优化

英文标题：Design and optimization on machining process for aircraft engine crankshaft based on neural network

作者：肖展开梅益罗宁康袁丽莉张然吴巧

分类号：TG316.3

出版年,卷(期):页码：2022,47(6):9-16

摘要：

以某发动机曲轴作为研究对象，通过DEFORM软件对曲轴预锻和终锻过程进行了金属流动和微观组织的耦合模拟，得出了不同工艺参数下曲轴锻造的模拟数值；采用Geomagic软件逆向工程技术，得到了最终处理完成后曲轴的实际尺寸，获得体积验证参数；最后，通过FORGE软件验证了其结果的正确性。建立了多输入、多输出的BP网络拓扑结构，以预锻温度、终锻温度、预锻阻尼墙高度、终锻阻尼墙高度、有无润滑、预锻速度和终锻速度等作为神经网络模型的输入参数，以终锻填充率、折叠量以及最大应力值性能指标作为神经网络模型的输出参数。通过神经网络预测值与正交试验下模拟值的对比，得出最优解下的工艺参数，发现训练后BP神经网络模型的相对误差较小，并具有很好的预测能力。

For an engine crankshaft, the coupled simulation of metal flow and microstructure was carried out on crankshaft pre\|forging and final forging processes by DEFORM software, and the simulation values of crankshaft forging under different process parameters were obtained. Then, the actual dimension of crankshaft after final treatment was obtained by Geomagic software reverse engineering technology, the volume verification parameters were obtained, and the correctness of the results was verified by FORGE software. Furthermore, a multi\|input and multi\|output BP network topology was established taking pre\|forging temperature, final forging temperature, pre\|forging damping wall height, final forging damping wall height, with or without lubrication, pre\|forging speed and final forging speed as input parameters of neural network model and performance indexes of final forging filling rate, folding amount and maximum stress value as output parameters of neural network model. Finally, the optimal process parameters were obtained by comparing the predicted values by neural network with simulated values under orthogonal test. The results show that the relative error of the trained BP neural network model is small and has good predictive ability.

基金项目：

贵州省科技计划项目（黔科合支撑[2019]2019）；贵州省科技支撑计划项目（黔科合支撑[2018]2175）

肖展开（1995-），男，硕士研究生 Email：xiaozhankai517@163.com 通信作者：梅益（1974-），男，博士，教授 Email：1141909227@qq.com

参考文献：

[1]陈志仁. 曲轴锻造过程金属流动与微观组织演化规律的数值模拟研究[D].济南：山东大学,2013.

Chen Z R.Numerical Simulation of Metal Flow and Microstructure Evolution in Crankshaft Forging Process [D].Jinan：Shandong University,2013.

[2]Zhao F, Zhang Z, Liu Y, et al. The coarse microstructure of mediumcarbon microalloyed steel crankshafts: Formation mechanism and finish forging control[J]. Steel Research International, 2020, 91(4):22-26.

[3]Zhang Z W, Li H, Zhou Y W, et al. Research on thermal deformation behavior of marine crankshaft S34MnV steel[J]. Steel Research International,2021,92(10):47-53.

[4]Guo C P, Cheng M B. Application of computer technology in optimal design of overall structure of special machinery[J]. Mathematical Problems in Engineering,2021, 3:258-267.

[5]Merayo D, RodríguezPrieto A, Camacho A M. Topological optimization of artificial neural networks to estimate mechanical properties in metal forming using machine learning[J]. Metals,2021,11(8):1289-1289.

[6]Bagliuk G A, Kyryliuk S F. Numerical analysis of the die forging of porous blanks in a die with the implementation of active friction forces[J]. Russian Journal of NonFerrous Metals,2021,62(1):11-16.

[7]莫建虎. 基于BP神经网络的挤压力与挤压出口温度预测研究[D].长沙：湖南大学,2009.

Mo J H. Prediction Research of Extrusion Pressure and Extrusion Outlet Temperature Based on BP Neural Network [D]. Changsha: Hunan University,2009.

[8]刘世平, 曹俊峰,孙涛,等.基于BP神经网络的冗余机械臂逆运动学分析[J].中国机械工程,2019,30(24):2974-2977,2985.

Liu S P, Cao J F, Sun T, et al. Inverse kinematics analysis of redundant manipulator based on bp neural network[J]. China Mechanical Engineering,2019,30(24):2974-2977,2985.

[9]Venet G, Baudouin C, Pondaven C, et al. Parameter identification of 42CrMo4 steel hot forging plastic flow behaviour using industrial upsetting presses and finite element simulations[J]. International Journal of Material Forming,2021,11(2)：17-21.

[10]黄礼琳, 高英俊, 邓芊芊, 等. 预变形条件下纳观裂纹扩展的微观机制研究[J].稀有金属,2020,44(1):63-71.

Huang L L, Gao Y J, Deng Q Q, et al. Study of microscopic mechanism of nanocrack propagation under predeformation[J]. Chinese Journal of Rare Metals， 2020,44(1):63-71.

[11]王铎. 一种杠杆式精密剪切模的虚拟设计[D]. 兰州: 兰州理工大学,2016.

Wang D. Virtual Design of a Lever Precision Searing Modie [D]. Lanzhou: Lanzhou University of Technology, 2016.

[12]Wang T, Ren Z K, He D P. Equivalent numerical algorithm for the striprolling process of a continuous variable crown mill using the coupled rigidplastic finite element method[J]. Journal of Marine Science and Technology,2019,27(2)：561-567.

[13]Kumar A M, Parameshwaran R, Rajasekar R. Simulation of 1.2765 DIN steel billet and crack analysis during the forging process using DEFORM-3D software[J]. Material Wissens Chaft und Werkstofftechnik， 2021,52(3)： 77-79.

[14]黄皓，付彭怀，彭立明，等模具温度和浇注温度对AZ91D镁合金热裂和流动性能的影响[J].特种铸造及有色合金,2012,32(1):81-84.

Huang H, Fu P H, Peng L M, et al. Effect of mold temperature and casting temperature on hot cracking and flow properties of AZ91D magnesium alloy [J].Special Casting and Nonferrous Alloy,2012,32(1):81-84.

[15]Bagliuk G A, Kyryliuk S F. Numerical analysis of the die forging of porous blanks in a die with the implementation of active friction forces[J]. Russian Journal of NonFerrous Metals,2021,62(1),21-25.

[16]Khasidashvili Z, Norman A J. Feature range analysis[J]. International Journal of Data Science and Analytics,2021,27(2)：33-38.

[17]张启森，彭桂枝，王晓莉. 基于正交试验的铝钢材料自冲铆接成形质量的多指标优化 [J]. 锻压技术，2020,45(3):87-91.

Zhang Q S，Peng G Z，Wang X L. Multiindex optimization on forming quality of selfpiercing riveting for aluminiumsteel materials based on orthogonal experiment [J]. Forging & Stamping Technology，2020,45(3):87-91.

[18]Wang Z Z, Adam Sobey. A comparative review between genetic algorithm use in composite optimisation and the stateoftheart in evolutionary computation[J]. Composite Structures,2020,233,237-251.

[19]Manoochehri M, Kolahan F. Integration of artificial neural network and simulated annealing algorithm to optimize deep drawing process[J]. Springer London,2014,73(1-4): 288-289.

[20]元莎，白玉冰，周乐育，等. 热处理工艺参数对Cr8冷作模具钢组织和性能的影响 [J]. 锻压技术，2020,45(1):168-172.

服务与反馈：

【本网站尚未开通全文下载服务】【加入收藏】