基于BP算法的大型船用曲轴成形载荷优化研究
Research and Optimization on Forming Load of Heavy Marine Crankshaft Based on BP Neural Networks
周 杰1， 朱若岭1， 徐戊矫1， 杜绍贵2
点击：3275次 下载：0次
作者单位：（1.重庆大学 材料科学与工程学院，重庆 400044；2.南车资阳机车有限公司，四川 资阳 641301）
中文关键字：大型曲轴； 有限元法； 正交实验法； 成形载荷分析； BP神经网络
英文关键字：heavy crankshaft; finite element; orthogonal design; forming load analysis; BP neural network
中文摘要：采用DEFORM-3D软件对新开发的大型全纤维曲轴TR镦锻过程进行数值仿真，得到该曲轴单拐的成形载荷。针对该型号曲轴单拐的成形载荷过大，导致设备过载的现象，利用正交实验法讨论了对锻件成形载荷有重大影响的三个关键工艺参数：模具与锻件的摩擦因子?滋、弯曲速度v和锻件的始锻温度T，得到了其对锻件成形载荷的影响规律。利用BP神经网络建立锻件最大成形载荷的计算模型，该模型比原来的有限元计算时间更短，提高了优化过程的效率。利用BP神经网络构建了主要影响因素对锻件最大成形载荷的响应面，并得到使成形载荷最优的设计参数范围。研究发现， ?滋和T的取值区间分别在0.2～0.4和1180～1200 ℃时，既能保证锻件顺利成形，又能降低成形载荷、保护设备。
英文摘要：The heavy crankshaft forging process was simulated by software DEFORM-3D, the forming load was obtained. To avoid forming overload in the heavy crankshaft, the key process parameters which influenced the forming load, such as friction factor ?滋, vertical velocity v and initial temperature T, were discussed by orthogonal experiment, and the influencing rules of above-mentioned factors on the maximal forming load were obtained. A computational model for the maximal forming load was presented based on BP neural Network. Compared with the traditional finite element model, the model can provide fast analysis and improve optimization efficiency. The response surface of the main factors to the maximal forming load was established by BP neural network and the optimal parameters of the forming load were obtained. The results show that the optimal values of ?滋 and T were within the range of 0.2～0.4 and 1180～1200 ℃, respectively, which can make sure the favorable die filling, decrease the forming load and protect forming equipment.