基于神经网络的铝合金压铸模热疲劳性能研究

Thermal Fatigue Property of Al-alloy Casting Die Based on Neural Network

建立了铝合金压铸模热疲劳性能的优化神经网络模型,并对该模型进行了试验验证。结果表明,该神经网络模型的相对训练误差和相对预测误差均小于5%。经神经网络模型优化后,3Cr2W8V铝合金压铸模的热疲劳裂纹级别从15级降为8级,H13钢铝合金压铸模从12级降为4级,表明两种材料的热疲劳性能均得到提高。

The neural network model on the thermal fatigue property of Al alloy casting die was established, and the model was examined by experiments. The results show that both the relative training error and prediction error are below 5%. And the crack grades of thermal fatigue for Al-alloy casting die of 3Cr2W8V and H13 decrease from 15 to 8 and from 12 to 4, respectively, after optimized by the neural network. It indicates that the thermal fatigue properties of the both materials are enhanced.