基于人工神经网络的水性涂料电脑测配色系统  

Water-borne Coatings Computer Color Control and Match System Based on Artificial Neural Networks

作　　者：杨涛[1] 廖宁放[1] 吴文敏[1] 吕大旻[1] 

机构地区：[1]北京理工大学颜色科学与工程实验室,100081

出　　处：《中国涂料》2006年第4期32-34,共3页China Coatings

基　　金：由北京市自然科学基金(No.4032016)支持。

摘　　要：介绍了基于多层BP人工神经网络的水性涂料配方预测模型的测配色系统,利用多层BP人工神经网络建立水性涂料反射样品色度参数空间到配方参数空间的非线性映射模型。对多种网络结构模型进行尝试,同时对通用千色卡(KEYTEC1190-TC)采集样本数据进行实验来讨论系统的可靠性,取得了满意效果。A system for paint color matching which is based on back-propagation (BP) neural networks is presented in this paper. The theory of the multi-layer BP neural networks model for the color recipe mapping is introduced, and the experiments for waterborne paints are carried out. A set of color chips in the KEYTEC 1190-TC color atlas was used to discuss the feasibility of the model. In addition, BP neural networks were trained to help the realization of the mapping between CIEXYZ space and recipes space for data generalization. The current results show that the networks performed well in the recipes mapping.

关 键 词：电脑配色 水性涂料 BP神经网络 

分 类 号：TQ630[化学工程—精细化工]