基于人工神经网络的7075高强铝合金反向挤压工艺构建及优化控制策略

Construction of optimization control strategy for indirect extrusion process of 7075 high-strength aluminium alloy based on artificial neural network

基于7075铝合金工业生产的反向等温挤压流程,通过有限元模拟和人工神经网络相互结合建立了数字工艺优化模型。首先,模拟了7075铝合金的反向挤压过程,模拟与试验的载荷误差不超过6.5%。随后对216组不同反向挤压工艺流程进行了数字建模,获得了挤压控制工艺、型材出口温度与成品性能的映射关系。将生成的反向等温挤压数据集用于构建反向传播神经网络。结果表明,该人工神经网络能精确预测反向挤压工艺参数与材料性能的关系,预测平均误差为0.83%,在此基础上建立了三维挤压成形极限图,为快速反向等温挤压质量控制提供了依据。

Based on the indirect isothermal extrusion process of 7075 aluminium alloy in industrial production,the digital process optimi-zation model was established by the combination of finite element simulation and artificial neural network.Firstly,the indirect extrusion process of 7075 aluminium alloy was simulated,and the error of load between simulation and test is less than 6.5%.Then,216 groups of different indirect extrusion processes were digitally modeled to obtain the mapping relationship between extrusion control process,profile outlet temperature and finished product performance.The generated indirect isothermal extrusion data set was used to construct the back propagation neural network.The results show that the artificial neural network can accurately predict the relationship between indirect ex-trusion process parameters and the material properties,and the average prediction error is 0.83%.On this basis,the three-dimensional extrusion forming limit diagram was established,which provided a basis for quality control of rapid indirect isothermal extrusion.