基于BP_GA算法的铸造有效应力预测与优化

为控制铸件凝固过程中的有效应力大小,避免热裂发生,利用有限元分析软件ProCAST对ZL205A铝合金牵引结构件低压铸造过程进行温度场模拟与有效应力预测,选择浇注温度、模具预热温度、传热系数和模具壁厚等影响铸造应力的因素作为设计参数。结合有效应力预测结果,构建4-7-1-1型神经网络和遗传算法以优化铸造工艺。结果表明,神经网络预测平均相对误差为1.45%,预测精度较高。通过遗传寻优方法,发现了最佳工艺参数组合:浇注温度为688℃,模具预热温度为291℃,模具壁厚为150mm,传热系数为1 284W/(m^2·K),并进行试验验证,获得品质较好的铸件。

Prediction and Optimization of Effective Stress in Casting Based on BP_GA Algorithm

In order to control the effective stress during the solidification process and avoid the occurrence of thermal cracking,the ProCAST software was used to simulate the temperature field and effective stress of the low pressure casting process of ZL205 Atraction component.Four factors,such as pouring temperature,mold preheat temperature,heat transfer coefficient and mold wall thickness were selected as design parameters.Combined with the results of effective stress prediction,a 4-7-1-1 artificial neural network and genetic algorithm were constructed to optimize the casting process.The results show that the average relative error of neural network prediction reaches 2.9%,presenting strong prediction accuracy.Through genetic optimization,the optimal process parameters are presented as follows:pouring temperature of 688℃,mold preheating temperature of 291℃,mold wall thickness of 150 mm,and the heat transfer coefficient of 1 284 W/(m^2·K).It provides reference and guidance for the low pressure casting process of traction structural parts.