GA-BP算法及其在冰铜品位神经网络模型中的应用  

Application of GA-BP to the Matte Grade Model Based on Neural Network

作　　者：汪金良[1] 卢宏[1] 曾青云[1] 

机构地区：[1]南方冶金学院,江西赣州341000

出　　处：《江西有色金属》2003年第3期39-42,共4页Jiangxi Nonferrous Metals

基　　金：江西省自然科学基金资助项目(编号 :0250026)

摘　　要：分析了遗传算法(GA)和BP算法的优缺点 ,将GA算法与BP算法有机地结合 ,形成了遗传BP算法。对冰铜品位神经网络模型的训练和检验结果比较表明 ,该算法优于BP算法 ,是一种快速全局搜索算法。模型训练结果也说明了神经网络技术应用于铜闪速熔炼过程建模的可行性和有效性。The advantages and disadvantages of Back Propagation Algorithm (BP) and Genetic Algorithm (GA) are analysed, combination of the two algorithms forms GA-BP. Compared the training and testing results of the matte grade neural network model with the GA-BP and BP respectively indicates that the GA-BP is superior to BP ,and it can search answer quickly in overall range. The training results also explains that nerual network technique is feasible and effective to the modeling for copper flash smelting process.

关 键 词：神经网络 遗传算法 BP算法 

分 类 号：TF811[冶金工程—有色金属冶金]