摘要
提出一种基于多层BP人工神经网络的粉末涂料配方预测模型;用BP算法人工神经网络建立粉末涂料反射样品的标准色度参数与配方浓度参数之间的映射关系。把人工神经网络的配方预测模型应用到典型的粉末涂料样品的测配色实验过程中。实验结果表明,基于多隐层BP网的模型可以实现粉末涂料样品的配方浓度空间与标准三刺激值颜色空间的相互映射,对64个节点的平均训练精度达到了1个CIELAB色差单位。
A recipe prediction model for color matching in powder paints production based on the BP neural networks is presented. The mapping between the colorimetric values and the recipe values in the reflective powder paints samples can be set up by the BP neural networks. The color matching experiments for typical powder paints are conducted by using such a model. The experimental results show that the mapping between the colorimetric space and the recipe space can be realized by the multi-layer BP neural networks, and the average prediction error for 64 training samples is less than 1 unit of CIELAB color difference.
出处
《光学技术》
EI
CAS
CSCD
北大核心
2005年第1期133-135,共3页
Optical Technique
基金
北京市自然科学基金(4032016)
北京理工大学基础研究基金(BIT_UBF_200301F16)
关键词
电脑配色
涂料颜色
配方预测
BP神经网络
computer color matching
paints color
recipe prediction
BP neural networks
