论文部分内容阅读
基于神经网络算法,以连杆材料、始锻温度、终锻温度、毛坯预热温度、模具加热温度和锻压速度为输入层参数,以屈服强度和耐磨损性能为输出层参数,构建了三层拓扑结构的连杆锻压工艺优化模型,并进行了模型的训练、预测和应用验证,以及连杆的屈服强度和耐磨损性能的测试与分析。结果表明,神经网络模型具有较强的预测能力和较高的预测精度。与生产线现用工艺相比,采用神经网络模型优化工艺制备的连杆屈服强度和耐磨损性能均得到明显提高;40Cr和42CrNiMo连杆屈服强度分别增加69 MPa、56 MPa,40Cr和42CrNiMo连杆磨损体积分别减小44%、40%。
Based on the neural network algorithm, the input rod parameters, the initial forging temperature, the final forging temperature, the preheating temperature of the blank, the heating temperature of the die and the forging speed are taken as the input layer parameters. The output layer parameters are the yield strength and the wear resistance, Layer topology of the link forging process optimization model, and the model training, prediction and application verification, as well as the connecting rod yield strength and wear resistance of the test and analysis. The results show that the neural network model has strong prediction ability and high prediction accuracy. Compared with the existing process, the yield strength and wear resistance of the connecting rod prepared by the neural network model optimization process were significantly improved. The yield strength of the 40Cr and 42CrNiMo connecting rods increased by 69 MPa, 56 MPa, 40Cr and 42CrNiMo respectively Wear volume decreased by 44%, 40%.