基于遗传神经网络的产品配色设计

Research on construction of color design system based on neural network and genetic algorithm

由于科技的进步和消费者意识的提高,加快了产品的生命周期,如何帮助设计师掌握消费者的感觉,提高产品创新的时效性,是一个不容忽视的问题.从感性工学的理论和知识出发,将目标产品(电话计费器)划分为色块的组合构成,以配色理论制作配色意象问卷,运用倒传递类神经网络学习的特性模拟配色意象评价模式,再通过遗传算法调适与更新的特性搜寻最符合需求意象的配色组合.研究结果显示:通过实验,依据类神经学习与遗传演算推演出的配色组合方式,最终可以架构系统而得以顺利执行,并且提供配色子代的样本作为建议.由本研究的结果可建立一个评估色彩感性的决策支持系统,藉以加快设计流程,辅助设计师以比较有效率且客观的方式进行产品色彩设计.

Due to advancement in science, development of technology and enhancement of customer＇s awareness, product lifecycle are greatly shortened and how to help designers master customer＇s feeling and escalate effectiveness of product innovation in time cannot be ignored. Starting from the theory and knowledge of Kansei engineering, object product, phone fee recoder, is divided into various color components. Then, a poll system designed by color combination theory is made. According to the theory of back-propagation neural network, customer＇s evaluation is simulated and finally, the optimal color combination through genetic algorithm iteration is gained. The results indicate that this system can carried out successfully with the abovementioned method. A decision support system for evaluation of color combination can be established by employing the results of this research to accelerate design flow. And assistant designer can perform product color design effectively and objectively.