【摘要】 在金属材料的热处理过程中，不同的热处理工艺参数会对材料的性能产生影响，然而热处理工艺参数的选择具有很强的经验性；对于高强高导铜合金，热处理工艺参数对其性能的影响更为显著。针对这一问题，采用改进的广义径向基函数(RBF)神经网络算法，对Cu-0.23Cr-0.2Zr-0.1V合金在热处理过程中固溶温度、固溶时间、时效温度和时效时间4组工艺参数下的合金电导率样本集进行训练、学习，建立了Cu-0.23Cr-0.2Zr-0.1V合金经时效处理后的电导率与固溶温度、固溶时间、时效温度和时效时间的映射模型。结果表明：采取广义RBF神经网络建立模型进行铜合金的时效性能预测是可行的，与传统的反向传播(BP)神经网络相比，广义RBF神经网络具有更高的输出精度和更好的泛化能力。更多还原

【Abstract】 In the process of heat treatment of metal materials, different heat treatment process parameters will affect the properties of materials, but the selection of heat treatment process parameters is very empirical. For high strength and high conductivity copper alloy, the influence of heat treatment parameters on its properties is more significant. Aiming at this problem, the generalized radial basis function(RBF) neural network algorithm was used to train and learn the electrical conductivity sample set of Cu-0.23 Cr-0.2 Zr-0.1 V alloy under four process parameters of solution temperature, solution time, aging temperature and aging time during heat treatment, and the mapping model between conductivity and solution temperature, solution time, aging temperature and aging time of the Cu-0.23 Cr-0.2 Zr-0.1 V alloy after aging treatment was established. The results show that it is feasible to use the generalized RBF neural network to establish a model to predict the aging properties of the Cu-0.23 Cr-0.2 Zr-0.1 V copper alloy. Compared with the traditional back-propagation(BP) neural network, the generalized RBF neural network has higher output accuracy and better generalization ability.更多还原