论文部分内容阅读
以凸轮式高速形变试验机得到的试验数据为基础,利用Matlab人工神经网络工具箱,建立了轴承钢的变形抗力与其化学成分、变形温度、变形速率及变形程度对应关系的RBF神经网络预测模型。分析了变形温度和变形速率对轧制压力网络模型精度的影响。得出随着变形温度的增加,网络的预测误差逐渐增大;随着变形速率的增大,网络的预测误差逐渐减小的结论。通过与BP网络和Elman网络模型相比较,结果表明,RBF网络模型具有更高的精度和较强的泛化能力。
Based on the experimental data obtained from the cam-type high-speed deformation testing machine, a RBF neural network prediction model of the deformation resistance of bearing steel and its chemical composition, deformation temperature, deformation rate and deformation degree was established by Matlab artificial neural network toolbox. The influence of deformation temperature and deformation rate on the accuracy of rolling pressure network model was analyzed. It is concluded that as the temperature of deformation increases, the prediction error of the network gradually increases. As the deformation rate increases, the prediction error of the network decreases gradually. Compared with BP network and Elman network model, the results show that RBF network model has higher accuracy and stronger generalization ability.