用灰关联分析法和人工神经网络优选达克罗涂液  

Optimization of Dacromet by Grey Relationship Analysis and BP Neural Network

作　　者：王哲[1] 刘记军[1] 阮宝科[1] 

机构地区：[1]解放军理工大学工程兵工程学院,江苏南京210007

出　　处：《材料保护》2004年第6期9-11,共3页Materials Protection

摘　　要：为寻找达克罗涂液配制的最优配比方案 ,基于灰关联分析方法 ,解析了达克罗涂液组分对涂层性能的影响。以CrO3 ,H3 BO3 ,锌粉和还原剂 4种组分作为输入数据 ,以涂层性能指标作为输出数据 ,采用 4 - 7- 2层结构 ,建立BP人工神经网络建立非线性模型 ,分别训练和预测了试验结果 ,同时以关联度的大小顺序为依据 ,分别拟合了各组分对涂层性能指标的影响曲线。根据预测结果 ,寻找出最优配比方案。采用BP人工神经网络结合灰关联分析方法 ,为达克罗涂液的制备提供了一种新的方法。To optimize the preparation formula of dacromet paint, the influence of dacromet paint components on the coat performance was studied by grey relationship analysis methods. The 4- 7- 2 nonlinear model was found by BP neural network with four components used as input data and the index of coat performance as output data, and the experimental results were trained and forecasted. Simultaneously, the influence curve of components on the coat performance was fittest in order of grey relational grade. The optimal formula is obtained according to the forecasted result. So the BP neural network combined with the grey relationship analysis provides a new way to prepare dacromet paint.

关 键 词：达克罗 灰关联分析 正交试验 BP人工神经网络 

分 类 号：TG174.4[金属学及工艺—金属表面处理]