节点文献

计算机辅助油漆配色系统的BP网络研究

【作者】 张全

【导师】 陆长德;

【作者基本信息】 西北工业大学 , 机械制造及其自动化, 2003, 硕士

【摘要】 BP网络是最具有代表性的人工神经网络之一,具有强大的数据处理能力和非线性映射能力。利用BP网络进行油漆配色,可以避免以KM理论、三光通理论为基础的传统方法需要处理大量复杂数据的缺点。本文就BP网络在计算机油漆配色中的应用作了深入研究:一、深入分析了油漆配色原理,以西安油漆厂提供的资料为例,对油漆色彩及其影响因素进行了量化和规范化处理,并建立了相应的BP网络模型;二、针对以往BP算法收敛速度慢、学习精度低的问题,提出了一种新的BP算法,应用结果证明新BP算法减少了局部极小问题与“假饱和”现象,提高了收敛速度、学习精度和泛化能力;三、开发了基于新BP算法的计算机辅助油漆配色原型系统,并结合西安油漆厂提供的资料对原型系统进行了测试,测试结果证明原型系统具有较强的学习能力和配色能力,达到了实用要求。

【Abstract】 The Back-Propagation (BP) network is one of the most useful Artificial Neural Network (ANN), which can deal with complicated data and nonlinear problem. Computer paint-color matching based on the BP network is a new method. This new method can avoid dealing complicated data, and the traditional methods based on Kubelka-Munk (KM) and 3-flux theory can’t. Having analyzed the principle of paint color matching, a new BP network model was constructed with the data provided by XI’AN Paint Company. However, the convergent speed and studying accuracy of the past BP algorithms are very low. A new BP algorithm was put forward in this paper. This new algorithm can improve the convergent speed, studying accuracy and endurance markedly. Finally, this paper developed a paint-color matching system based on the new BP algorithm. The paint-color matching system can carry out studying and color matching very well, thus it has the value of practicality.

  • 【分类号】TP183
  • 【被引频次】5
  • 【下载频次】130
节点文献中: 

本文链接的文献网络图示:

本文的引文网络