木材染色颜料配方预测模型  

Predictive Model of Wood Dyeing Pigment Formula

作　　者：管雪梅[1] 郭明辉[2] 曹军[1] 

机构地区：[1]东北林业大学机电工程学院,哈尔滨150040 [2]东北林业大学,东北林业大学生物质材料科学与技术教育部重点实验室,哈尔滨150040

出　　处：《科技导报》2013年第17期29-32,共4页Science & Technology Review

基　　金：黑龙江省自然科学基金项目(C201127);中央高校基本科研业务费专项资金项目(DL11BB24)

摘　　要：木材颜色是决定消费者印象的重要因素,为了提高木制品的装饰作用和产品价值,要对木材及木质材料进行着色。将计算机配色的方法用于木材染色中,能加快染料配方生成的速度并将极大地提高工作效率、节约成本。本文研究了一种运用动态模糊神经网络建立的木材染色颜料配方预测模型,所谓的"动态"是指模糊神经网络的网络结构不是预先设定的,而是动态变化的,即在学习开始前,没有一条模糊规则,其模糊规则在学习过程中逐渐增长而形成的。在论述建模方法的基础上,对算法中的学习规则和参数确定进行了研究。模型为三输入三输出系统,输出就确定为活性艳红X-3B、活性黄X-R、活性蓝X-R的浓度值,输入为色差,该模型预测相对误差为0.52%,训练时间为128s,结果比较令人满意。这种方法为木材染色配色提供了一种新的途径,同时也为其理论在配色系统中的应用提供了新的思路,具有一定的理论研究价值和实际应用价值。Wood color is an important factor determining consumer first impression. In order to improve the decorative role and value of wood products, wood and wood materials need to be coloring. Applying computer color matching method to wood dyeing for speeding up the generation of dyeing formula would greatly improve work efficiency and save costs. A kind of prediction model for wood dyeing pigment formula is built by using Dynamic Fuzzy Neural Network （DFNN）. The word ＂dynamic＂ refers to the fact that the network structure of fuzzy neural network does not preset, it changes dynamically; that is to say, there is no predeterminate fuzzy rule before learning, its fuzzy rules gradually increase and form during the learning process. The output is concentration values of reactive brilliant red X-3B, reactive yellow X-R, and reactive blue X-R, input is color difference, namely, △L, △a, and △b. The relative error of the prediction model is 0.52% and its training time is 128s. The results are comparatively satisfactory. The method provides a new way for wood dyeing and color matching and a new idea for the applications.

关 键 词：计算机配色 动态模糊神经网络 预测模型 

分 类 号：S781[农业科学—木材科学与技术]