基于遗传神经网络的等离子喷涂纳米ZrO_2-7%Y_2O_3涂层工艺参数优化

Process parameters optimization of nanostructured ZrO_2-7%Y_2O_3 coating deposited by plasma spraying based on genetic algorithms and neural networks

将BP神经网络和遗传算法相结合用于等离子喷涂纳米ZrO2-7%Y2O3涂层的工艺参数优化,根据正交试验结果对模型结构进行训练,建立了喷涂距离、喷涂电流、主气压力、辅气压力与涂层结合强度和显微硬度之间的BP神经网络模型,并基于遗传算法对涂层结合强度和显微硬度进行了单目标和多目标参数优化.结果表明,模型预测值与试验值十分接近,说明该网络模型是正确和可靠的.遗传算法优化的涂层最大结合强度和显微硬度(HV)分别为44.0 MPa和12.663 GPa;当涂层结合强度和显微硬度两个性能参数权重相同时,在喷涂距离90.66 mm、喷涂电流934.63 A、主气压力0.304MPa和辅气压力0.898 MPa时涂层综合性能最优.

BP neural networks and genetic algorithms were combined to optimize process parameters of the nanostructured ZrO2-7%Y2O3 coating prepared by plasma spraying technique.The neural networks were trained based on the experimental results of orthogonal tests,and the BP neural networks model was developed to describe the relationship between coating properties(bonding strength and microhardness) and four main process parameters,including spraying distance,spraying electric current,primary gas pressure and secondary gas pressure.Meanwhile,the bonding strength and microhardness of the nanostructured coating were optimized with single-objective and multi-objective optimization methods based on the genetic algorithms.The results show that the prediction data agrees well with the experimental values,which indicates that the proposed model is correct and reliable.The maximum bonding strength and microhardness of the coating are 44 MPa and 1 266 HV,respectively.The overall performance of the coating is best when the spraying distance is 90.66 mm,spraying electric current 934.63 A,primary gas pressure 0.302 MPa and secondary gas pressure 0.892 MPa while keeping the weight of bonding strength and microhardness constant.