论文部分内容阅读
采用5×25×2三层拓扑结构,以始锻温度、终锻温度、锻压速度、模具预热温度、模具预热时间为输入层参数,以室温耐磨损性能和高温耐磨损性能为输出层参数,构建出前轴锻压工艺优化的神经网络模型。结果表明,神经网络模型的预测能力较强,预测精度较高。与生产线传统工艺相比,采用神经网络模型优化工艺制备的前轴室温和高温磨损体积分别减小了39%和42%,室温和高温耐磨损性能均得到明显提高。
The 5 × 25 × 2 three-layer topology was used as input layer parameters to initial forging temperature, final forging temperature, forging speed, mold preheating temperature and mold preheating time. The wear resistance at room temperature and high temperature wear resistance were Output layer parameters to build a front-axle forging process optimization neural network model. The results show that the neural network model has stronger prediction ability and higher prediction accuracy. Compared with the traditional process of the production line, the wear volume of the mild and high temperature of the front axle chamber prepared by the neural network model optimization process is reduced by 39% and 42% respectively, and the wear resistance at room temperature and high temperature are obviously improved.