摘要
用传统的数学建模方法很难根据工艺参数对材料力学性能进行预测,而BP神经网络存在预测精度不高等缺点。为此,本文采用适应性更好的RBF(径向基函数)神经网络在其它工艺参数不变的条件下,根据铸轧速度和浇注温度两个关键参数对7050铸轧件的抗拉强度、屈服强度和伸长率进行预测。拉伸试验和Matlab仿真的结果表明:本文构建的RBF神经网络的预测精度高于同条件下的BP神经网络,非常接近于试验值,具有较高的预测精度。
The traditional mathematical modeling method is difficult to predict the mechanical properties of materials according to the process parameters,while the BP neural network has the disadvantages of low prediction precision. Therefore,under the other technological parameters as constant,the tensile strength,yield strength and elongation of 7050 casting rolling parts were predicted by using RBF (radial basis function)neural network with better adaptability according to the two key parameters of casting speed and pouting temperature.The results of tensile test and Matlab simulation show that the prediction accuracy of the RBF neural network constructed in this paper is higher than that of BP neural network under the same conditions,which is very close to the experimental value and has a high prediction accuracy.
作者
苏燕
梁武
SU YAN;LIANG WU(Beihai Vocational College,Beihai 536000,China)
出处
《热加工工艺》
CSCD
北大核心
2018年第21期145-147,151,共4页
Hot Working Technology
基金
广西职业教育教学改革立项项目(GXGZJG2015B008)
关键词
7050铝合金
铸轧
RBF神经网络
力学性能
7050 aluminum alloy
rolling casting
RBF neural network
mechanical properties